# Pathwise Academy — Full content

> Companion to https://pathwiseacademy.co/llms.txt. Full Markdown body of every blog post inlined for LLM ingestion. See https://pathwiseacademy.co/llms.txt for the site index + key facts.

Pathwise Academy runs hands-on AI classes for kids (10–17), adults, companies, and schools across Singapore, Hong Kong, and APAC. Every learner walks out with something real they built using AI tools — a deployed web app on the coding track, or an AI-animated video, cloned voice and podcast episode on the creator track.

Contact: hello@pathwiseacademy.co · WhatsApp +65 8026 1562 · https://pathwiseacademy.co

---

## ChatGPT for schools in Singapore: what leadership actually needs to know

> Where Singapore schools stand on ChatGPT and generative AI in 2026 — the shift from caution to structured guidance, what 'well-run' actually looks like, and the questions your SLT will ask.

Published: 2026-07-09 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/chatgpt-for-schools-singapore

Singapore schools have largely moved past the "ban it or ignore it" phase on ChatGPT and generative AI — the more useful question for school leadership now is what a well-run programme looks like, not whether to allow it. The version that works is **build-based and supervised**: students use the tool to make something real, with an adult in the room and clear framing, rather than unstructured open access. The lowest-risk way to start is a single contained pilot — one year group, one day. If you want the fuller future-skills case first, see [why international schools should teach AI](/blog/why-international-schools-should-teach-ai); if you're ready to pitch it, see [how to pitch an AI hackathon to your school leadership](/blog/how-to-pitch-ai-hackathon-to-school-leadership).


A curriculum lead at a Singapore international school asked me this at the start of a discovery call, almost apologetically:

> "I feel like we're behind. Every other school seems to have 'sorted out' their ChatGPT policy already, and we're still arguing about it internally. Have we missed the boat?"

No. Most schools are somewhere in the middle of this conversation, not at the end of it. Here's the honest, current picture for Singapore school leadership specifically — where things stand, what a well-run programme actually looks like, and the questions your SLT will ask before signing off.

## Where things actually stand in Singapore

The broad direction of travel in Singapore has been from caution toward structured guidance: rather than treating ChatGPT and generative AI as something to keep out of the classroom, the emphasis has shifted to teaching students to use it responsibly, with appropriate safeguards, under supervision. That's the system-level direction. What it means for your specific school depends on your own device policy, your board, and your own risk appetite — which is exactly why this is a leadership decision, not something that resolves itself.

The practical implication: if your school is still operating on a blanket ban written a couple of years ago, it's very likely out of step with where the wider system and most peer schools have already moved. That doesn't mean opening the floodgates — it means the conversation has shifted from *whether* to *how well*.

## ChatGPT specifically, or "generative AI" generally?

Worth naming directly: ChatGPT is the tool most parents and staff have heard of, so it's often used as shorthand for the whole category. In practice, the specific tool matters far less than the format. Claude, ChatGPT, and other frontier tools are all capable of the same core thing — a student describes what they want and the AI helps produce a first draft, a piece of code, a design — and the differences between them matter less to a school than *how* the tool is being used.

The question worth asking isn't "should we allow ChatGPT?" It's "what are students actually doing with whichever tool we allow?" A student using ChatGPT to skip the thinking and hand in the output is a real problem. A student using ChatGPT (or Claude, or any equivalent) as a supervised drafting partner, in a structured session where the deliverable is something they built and can explain, is a genuinely good use of class time. Same tool category, opposite outcome.

## What "well-run" actually looks like


**What separates a well-run ChatGPT/AI programme from a risky one:**

- Students *build and produce something real* — not open-ended, unsupervised chat access
- An adult is in the room, and outputs are visible to a teacher, not hidden in a personal device
- The session has a clear brief: what students are meant to produce and how they'll explain it
- Small enough groups, or enough mentor support, that a teacher can see what's actually happening on screen
- The teaching targets judgement — spotting when the AI is wrong, editing critically — not just "how to prompt"
- It's framed as directed practice, not vague "AI exposure"


This is the same standard we'd apply to any tool with real capability and real risk if misused — the classroom management principle isn't new, only the tool is.

## The questions your SLT will actually ask

In rough order of how often they come up:

### "Won't this just make cheating easier?"

The honest answer: unsupervised, yes, it can. Supervised and structured, it does the opposite — students editing against a fluent AI-generated draft tend to get *more* critical of writing, not less, because they're now evaluating something rather than staring at a blank page. The risk lives in the unsupervised version, which is exactly the version a well-run in-class programme avoids.

### "Is it safe for younger students?"

Use tools with content guardrails, keep an instructor or teacher actively monitoring outputs during the session, and be ready to walk your safeguarding lead through the approach before anything starts. The materially less safe path is students discovering ChatGPT unsupervised on their own devices at home, with no one having taught them how to use it well — which is happening at most schools already, guidance or not.

### "Will our teachers need to become AI experts overnight?"

No, if you bring in instructors who lead the technical side. Teachers keep doing what they're already good at — watching the room, supporting students who are stuck or checked out — while a specialist runs the technical content. Most teachers report learning more about the tools by walking the room during a session than they would from a training day.

### "What's the lowest-risk way to actually start?"

A single, contained pilot: one year group, one day, a clear start and end, no permanent timetable change. It produces visible results the same afternoon, and gives you real evidence before deciding whether to scale further. The full playbook for proposing this to your SLT is here: [how to pitch an AI hackathon to your school leadership](/blog/how-to-pitch-ai-hackathon-to-school-leadership).

## The school offering, at a glance



## Common questions

**Is ChatGPT allowed in Singapore schools?**
Singapore's education system has broadly moved from caution to structured guidance on generative AI, with schools expected to teach responsible, supervised use rather than ban it outright. Individual schools still set their own device and app policies, so the honest answer is: increasingly yes, but check your specific school's current policy rather than assuming a blanket rule either way.

**Should a Singapore school use ChatGPT specifically, or another AI tool?**
ChatGPT is the most recognised name, but it's one of several capable tools (Claude among them) — the tool matters less than whether the programme is build-based, supervised, and produces something real students made, rather than open, unstructured chat access with no framing.

**What's the safest way to introduce ChatGPT or generative AI at a Singapore school?**
Start with a single, contained pilot — one year group, one day, a hackathon format with a fixed start and end — rather than opening AI access school-wide at once. It's low-risk, produces visible results the same day, and gives you real evidence before deciding whether to scale.

**Will using ChatGPT in class stop students learning to write and think for themselves?**
Only if it's used to produce the final output unsupervised. Used well — as a drafting partner students then critique, edit, and defend — it raises the bar on what students notice about their own writing, because they're now editing against a fluent baseline instead of a blank page.

## Where to start

If you want the broader future-skills argument first, read [why international schools should teach AI](/blog/why-international-schools-should-teach-ai). If you're ready to build the proposal for your own SLT, [how to pitch an AI hackathon to your school leadership](/blog/how-to-pitch-ai-hackathon-to-school-leadership) has the one-page format that tends to get a yes.

Or skip straight to a [discovery call](/schools) — we'll ask about your priorities and your worries, and follow up with a written proposal a few days later. No slides on that call, and no pressure either.

You don't need to have ChatGPT policy fully resolved before you start. You need one contained, well-run pilot to find out what your students actually do with it.

— *Mr. Brown*


---

## Claude Code for kids: can children actually use it? A teacher's guide

> Yes — kids as young as 10 can use Claude Code to build real, deployed apps. Here's how it actually works for a child, what a first project looks like, and how to set it up safely.

Published: 2026-07-09 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/claude-code-for-kids

**Short answer: yes.** Claude Code is genuinely usable by kids — not "usable with heavy help," actually usable, by the child, as the one doing the building. The reason is simple: it removes the part that used to gatekeep kids out of coding (syntax, semicolons, terminal errors) and leaves the part kids are naturally good at (describing an idea clearly and iterating on feedback). Ages **10+** is where we see it click; **11–14** is the sweet spot in our classes. It's not a kids' app, so an adult should set up the account — but once it's running, most 10-year-olds need less hand-holding than their parents expect.


A parent asked me this at pickup last week: *"My son keeps saying he 'built an app with Claude Code' this weekend. Is that actually true, or is he exaggerating?"*

It was true. He'd built a small tool that quizzes him on Spanish vocab from a list he typed in, and it was sitting on a real URL he could send his friends. He's 11. He'd never written a line of code before that Saturday.

Here's the honest, non-marketing version of how that's possible — and how to set it up for your own kid.

## What Claude Code actually is (in one paragraph)

Claude Code is Anthropic's AI coding agent. Instead of giving you code to copy and paste, it writes the code, runs it, checks whether it worked, fixes what's broken, and keeps going until the thing you described actually exists. You talk to it in plain English — "make the button bigger," "the quiz should show one question at a time" — and it edits the real files itself. We've written the [full plain-English primer here](/blog/what-is-claude-code) if you want the deeper version; this post is specifically about how it plays out for a child.

## Why it works for kids specifically

The old barrier to a kid coding wasn't creativity — kids have plenty of ideas. It was syntax. A missing semicolon, a misspelled variable name, an indentation error, and the whole thing breaks in a way that's opaque to a beginner and genuinely no fun to debug. That's where most kids gave up, usually inside the first hour.

Claude Code removes that barrier entirely. The child never has to get the syntax right — they have to get the *description* right, and then react to what comes back. That's a skill kids are often better at than adults: they're less precious about a first draft, they iterate fast, and they're blunt about what looks bad ("that's ugly, make it bigger") in a way that's exactly the right instinct for directing an AI builder.

What's left for the child to actually learn is the good stuff:
- **Breaking a big idea into a first small piece** ("let's get one question working before we add ten")
- **Describing precisely what they want** ("bad" is copyable, "the text needs to be blue and centered" is buildable)
- **Reading output and reacting** — does this match what I asked for? What's still wrong?

That's closer to product thinking than programming, and it's a skill that transfers no matter what a child does later.

## What a first session actually looks like

Concrete, not hypothetical — this is roughly how it goes for a 10-13 year old with zero prior coding experience:

1. **Pick one small, specific thing.** Not "a game." More like "a game where you click the right animal before the timer runs out." Specific and small beats big and vague every time.
2. **Describe it like a brief, not a wish.** "I want a page with a countdown timer and pictures of animals. When you click the right one before time runs out you get a point." That's enough for Claude Code to build a working first version.
3. **Look at what it built and react.** This is where kids come alive — "the timer's too fast," "I want it to say 'nice!' when I win," "can the background be space instead of white?" Each round, Claude Code edits the actual files and it just... changes.
4. **Ship it.** By the end of the day, there's a real URL. Not a mockup — a live link the child can send to a grandparent.

The pattern holds across ages: a homework quiz generator, a habit tracker, a personal website, a small game. The build is real software, deployed, with the child's name on it.

## Setting it up for a child (the practical bit)

Claude Code is a developer tool first — it wasn't designed with a 10-year-old as the primary user, which means the setup step needs an adult, even though the *building* step doesn't.

1. **Account in a parent's name.** Anthropic's terms require account holders to be 18+, so set it up under yours and let your child use it with you, same as you'd do for [Claude in general](/blog/is-claude-safe-for-my-10-year-old).
2. **Expect to help with installation, once.** Claude Code runs as a terminal app. Getting it installed is a five-minute, slightly technical step — worth doing together the first time rather than handing a child a set of instructions cold.
3. **Sit in for the first session.** Not because the content is risky — it's because the first stuck moment (an install hiccup, a confusing error) is exactly where an unsupervised kid gives up. One adult in the room for session one usually means the child is fully independent by session two.
4. **Start tiny.** The single biggest predictor of a good first session is scope. "A tiny tool that does one thing" beats "an app" every time — big ideas are fine for session three, not session one.

If you want the deeper safety conversation — content safety, privacy, the cognitive-offload question parents actually worry about — [that's covered in full here](/blog/is-claude-safe-for-my-10-year-old). Everything in that post applies to Claude Code too; it's the same underlying model, just pointed at a keyboard instead of a chat box.

## "But is this really coding, or just prompting?"

Fair question, and worth answering directly rather than dodging it. It's closer to *directing* than to *typing syntax* — which is exactly why it's sometimes called ["vibe coding."](/blog/what-is-vibe-coding) The child isn't learning where semicolons go. They're learning to specify an outcome clearly, evaluate whether the result matches, and give precise feedback until it does — which is arguably a more durable and more transferable skill than memorising a language's syntax, especially given how fast the tools themselves keep changing.

It's also not "the AI did it for them" in the way a parent might fear. A vague brief gets a mediocre, generic result. A clear, specific, opinionated brief — the kind that comes from a kid who actually cares what their game looks like — gets something genuinely good. The taste and the judgment are still entirely the child's.

## How we teach this at Pathwise

Claude Code is the primary tool in our **Idea to App** program for [young builders](/young-builders) (ages 10–17). Cohorts run at ~8 students so every child gets real attention, taught by working teachers who build software themselves — not just classroom facilitators reading a script. Every student leaves with a real, deployed project and a link they can actually share.

If you want to see it happen in person before deciding — [browse the next cohort dates](/young-builders#schedule) or [have a look at the young builder programs](/young-builders).

The hardest part for most kids isn't Claude Code. It's deciding what they want to build. Once they've got an idea they actually care about, the tool gets out of the way faster than most parents expect.

— *Mr. Brown*


---

## Claude Code in Singapore: where to learn it and who's already using it

> Claude Code has quietly become the tool Singapore founders, marketers, students and even schools use to build real software without hiring an engineer. Here's the local picture — who's using it, and where to learn it in person.

Published: 2026-07-09 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/claude-code-in-singapore

**Claude Code has landed in Singapore** — not as a niche developer tool, but as the thing founders, marketers, students, and a growing number of schools are using to build real software without hiring an engineer. If you want the plain-English explainer first, read [what Claude Code actually is](/blog/what-is-claude-code). If you want to learn it hands-on in Singapore, small in-person cohorts exist for both [young builders](/young-builders) (10–17) and [adults](/adults), run at JustCo Marina Square and online — every learner leaves with a deployed app.


A founder I met at a coffee shop near Marina Square put it well:

> "Six months ago, 'building a prototype' meant a two-week Upwork search. Now I just open Claude Code on a Sunday and have something live by Monday."

That's not a Silicon Valley story. That's Singapore, right now. Here's the honest local picture — who's actually using Claude Code here, and where to learn it if you want in.

## What Claude Code is, in one line

Claude Code is Anthropic's AI coding agent — you describe what you want in plain English, and it writes, runs, tests, and fixes the code itself, rather than handing you a snippet to copy and paste. For the full plain-English primer, see [what is Claude Code?](/blog/what-is-claude-code)

## Who's using Claude Code in Singapore right now

Four groups, based on what I'm actually seeing in classes and conversations here:

### 1. Founders and solo operators

Singapore's founder density is high, and the "build the prototype myself before hiring an engineer" move has become the default rather than the exception. Claude Code is usually the tool doing the building — landing pages, internal dashboards, first-version MVPs.

### 2. Marketers and ops people

The person who used to file a ticket with engineering and wait two sprints is now building the internal tool themselves on a Sunday. Stripe dashboards, CRM automations, one-off internal calculators — the kind of thing that was never important enough to be an engineering priority, but was always annoying to do by hand.

### 3. Students, from secondary school through university

We're seeing this from both directions: NUS/NTU students using Claude Code for portfolio projects and hackathons, and — more surprising to most parents — secondary school students building real, deployed apps. If you're wondering whether a child can genuinely use it (not "with heavy help," actually use it), see [Claude Code for kids](/blog/claude-code-for-kids).

### 4. Schools running hackathons

A number of Singapore international schools have started running whole-school AI hackathon days, with Claude Code as the primary build tool for the afternoon. More on why that's a better instinct than banning AI outright: [why international schools should teach AI](/blog/why-international-schools-should-teach-ai).

## Why Singapore specifically is a good place to pick this up

A few local factors line up in Claude Code's favour here:

- **English-medium by default**, which is Claude Code's native working language — no translation layer between what you mean and what you type.
- **A dense, compact founder and SME scene** — Singapore's small enough that "the person three tables over already tried this" is often true, which speeds up word-of-mouth adoption.
- **A schools sector already comfortable with structured after-school enrichment** — the infrastructure for "kids learn a serious skill on a Saturday" already exists here; AI building slotted into it faster than in markets without that culture.

## Where to actually learn it in Singapore

Two honest paths, in order of commitment:

### 1. Teach yourself this weekend

Get an Anthropic account at [claude.com](https://claude.com), install Claude Code, and build one small, specific thing — not "an app," more like "a one-page tool that does this one annoying thing I keep doing by hand." The full three-step starter path is in the [Claude Code primer](/blog/what-is-claude-code).

### 2. Learn it in a small, in-person cohort

If you want a teacher in the room rather than trial and error alone, we teach Claude Code as the primary tool in our **Idea to App** programme:

- **Ages 10–17** — [young builders classes](/young-builders), after-school and weekend cohorts
- **Adults** — [adult classes](/adults), non-coders welcome, same build-based format

Both run in person at JustCo Marina Square (City Hall MRT) and online for anyone in Singapore. Small cohorts (~8 people), real professional tools, taught by working teachers who build with Claude Code themselves — not slide decks. See the [Singapore programme page](/singapore) for the full picture, or [browse upcoming dates](/classes).

## Common questions

**Is Claude Code used in Singapore?**
Yes — increasingly by founders building first prototypes, marketers and ops people automating their own workflows, students building portfolio projects, and a growing number of schools running AI hackathons. It's the same tool globally; what's local is who's teaching it and where you can learn it in person.

**Where can I learn Claude Code in Singapore?**
You can teach yourself from Anthropic's own docs, or learn it in a small, in-person cohort. Pathwise runs Claude Code as the primary tool in its Idea to App programme for both young builders (10–17) and adults, in person at JustCo Marina Square and online for anyone in Singapore.

**Do I need to already know how to code to use Claude Code in Singapore classes?**
No. Every Pathwise cohort — kids and adults — assumes zero prior coding background. You describe what you want in plain English; Claude Code writes, runs, and fixes the code. The skill being taught is clear direction, not syntax.

**Is there a Claude Code community or meetup in Singapore?**
Singapore's broader AI-builder and vibe-coding meetup scene (Slack groups, Luma events, hackathons) increasingly features Claude Code users, though there isn't yet a dedicated Claude Code-only meetup. The fastest way to meet other local users in person is a hands-on class or hackathon.

## Pathwise in Singapore — at a glance



## Where to go from here

If you're not ready to commit to a class, start with the [Claude Code primer](/blog/what-is-claude-code) and build one tiny thing this weekend. If you'd rather learn with a teacher in the room, [browse the Singapore programme](/singapore) or [see upcoming cohort dates](/classes).

The tool is the easy part. Deciding what to build with it is where most people actually get stuck.

— *Mr. Brown*


---

## How can parents use AI? A practical guide beyond homework help

> Most parents only meet AI through their kid's homework. Here are the actual uses that save time and reduce mental load — plus where the line is with helping your child's schoolwork.

Published: 2026-07-09 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/how-can-parents-use-ai

Most parents' only contact with AI is through their kid's homework — which means they miss the uses that actually save *them* time. The highest-value ones: drafting awkward emails, planning the week (meals, schedules, logistics), summarising long documents before committing to read them, and thinking through a decision by talking it out with AI first. None of it requires coding. If you also want the kid-facing side of this — introducing AI to your child well — that's a separate guide: [how to introduce AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old).


A mother in one of our parent info sessions said something that stuck with me:

> "I keep hearing about AI because of my son's homework. I've genuinely never used it for myself. What would I even use it for?"

That's a more common position than it looks. A lot of parents' entire AI exposure runs through their child — watching them build with it in a class, or worrying about whether ChatGPT wrote the essay. Meanwhile the parent's own week — the emails, the meal planning, the school forms, the sheer mental load of running a household — goes untouched.

Here's the practical version: what AI is actually good for in a parent's day, separate from anything to do with your kid's schoolwork.

## The mental-load uses (the biggest win, and the least talked about)

Parenting runs on a huge amount of small, low-stakes-but-annoying cognitive labour. This is where AI earns its keep fastest.

- **Drafting the awkward email.** The one to the teacher about the seating change, the one to another parent about a birthday party mix-up, the one to your boss asking for flexibility around a school event. Describe the situation and what you want the outcome to be; let AI draft it; edit for your voice. Ten minutes becomes two.
- **Planning the week's meals around what's already in the fridge.** List what you have, describe your family's constraints (a fussy eater, a nut allergy, a Tuesday night that's always chaotic), and get a plan that isn't "order takeout again."
- **Turning a messy week into a clear schedule.** Paste in three separate WhatsApp threads about pickup times, a birthday party, and a dentist appointment, and ask for one clean list. This alone is worth trying once — it's genuinely satisfying to watch chaos become a list.
- **Summarising long documents before you commit to reading them.** A school's new phone policy PDF, an insurance renewal, a tenancy agreement. Paste it in, ask "what actually changed, and what do I need to do?" You can always read the full thing after — but now you know whether you need to.

None of this requires any technical skill. It's a chat window and a clearly described situation.

## The thinking-partner uses

A second category, slightly less obvious but just as useful:

- **Talking through a parenting decision you're stuck on.** Not to outsource the judgement — the judgement stays yours — but to get a second set of considerations you hadn't thought of. "My 13-year-old wants a phone. What are the actual tradeoffs I should be weighing, not the generic ones?"
- **Getting a second opinion before a hard conversation.** With a co-parent, a teacher, your own parent about grandparenting boundaries. Describe the situation honestly (including your own part in it) and ask what you might be missing.
- **Understanding something your kid is into, fast.** Your child is deep into a game, a YouTuber, a hobby you don't understand. Five minutes of "explain this to me like I'm a curious adult, not a kid" gets you enough context to have a real conversation with them about it — which matters more to a teenager than most parents realise.

## Where the line is: your kid's homework

This is the question every parent eventually asks, so let's answer it directly rather than dodging it.

**Using AI to understand a topic yourself, so you can explain it well, is fine — and often better than fine.** If your child is stuck on a maths concept and you've forgotten how it works, ask AI to re-teach *you*, then you teach your child. That's using AI the way you'd use a textbook.

**Using AI to produce the answer your child then copies is where it stops being fine.** It costs your child the actual learning, and it's usually more visible to a teacher than a parent expects — a piece of writing that doesn't sound like your child's writing is not subtle.

The test I give parents: *could your child explain, in their own words, what got handed in?* If yes, however AI was involved in getting there, you're fine. If no, something needs to change.

For the fuller version of this — how to introduce AI to your child well, not just police it — see [how to introduce AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old).

## If you want to go further than chatting

Everything above uses AI as a chat tool — no building, no code. Some parents, once they've gotten comfortable with that, want to go a step further: building an actual small tool for their family (a chore tracker, a shared meal planner, something specific to how their household runs). That's a different skill — closer to what we teach in adult classes — and it's optional, not a requirement for using AI well as a parent. If you're curious, the [Claude Code primer](/blog/what-is-claude-code) is the place to start, and [adult classes](/adults) exist if you want to learn it properly with a teacher in the room.

## Common questions

**How can parents use AI day to day?**
The highest-value everyday uses are the unglamorous ones: drafting the awkward school email, planning a week of dinners around what's already in the fridge, turning a chaotic family week into a clear schedule, and summarising a long document (permission slip, insurance policy, tenancy agreement) before you commit time to reading it fully.

**Is it okay for a parent to use AI to help with their child's homework?**
Using AI to understand a topic yourself so you can explain it well is fine. Using it to produce the answer your child then copies isn't — that costs your child the actual learning, and most teachers can tell. The test: could your child explain what got handed in?

**Do parents need to learn to code to use AI usefully?**
No. Almost everything in this guide uses AI chat tools (Claude, ChatGPT) with no code involved. Building an actual tool — a family chore tracker, a small automation — is a separate, optional step for parents who want to go further, not a requirement for everyday use.

**What's the best AI tool for a busy parent to start with?**
Claude or ChatGPT, used as a chat tool, covers most of what's in this guide. Start with one real task you're avoiding this week — an email, a plan, a summary — rather than trying to learn the tool in the abstract.

## Where to go from here

Two directions, depending on what brought you here:

1. **For yourself** — pick one thing from this guide (the awkward email, the meal plan) and try it this week before doing anything else.
2. **For your child** — if you're here because of your kid, not yourself, the guide you actually want is [how to introduce AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old). And if they're ready to go deeper than chatting — building a real, deployed app — we run [classes for ages 10–17](/young-builders) in Singapore, Hong Kong, and online.

The AI doesn't need to be a mystery your kid understands and you don't. It's useful for you too — just for different, more boring, more useful things.

— *Mr. Brown*


---

## Learning AI for kids: Singapore vs Hong Kong, compared

> How AI education for kids actually differs between Singapore and Hong Kong — market maturity, pricing, format, language of instruction — and what's identical no matter which city you're in.

Published: 2026-07-09 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/learning-ai-for-kids-singapore-hong-kong

Singapore and Hong Kong both have real, build-based AI education for kids (10–17) now — same core format: small cohorts, real professional tools, every child ships a live app. The differences are market maturity (Singapore's is more crowded), pricing (Hong Kong runs roughly HKD 350–600/session online vs Singapore's SGD 60–100, which converts to a similar real cost), and how much STEM-tutoring rebranding you have to filter through (more of it in Hong Kong). If you want the full city-specific breakdown, read the [Singapore guide](/blog/ai-classes-for-kids-singapore) or the [Hong Kong guide](/blog/ai-classes-for-kids-hong-kong) directly — this post is for comparing the two.


A parent who splits her year between Singapore and Hong Kong asked me something I hadn't been asked before:

> "Is what my daughter would learn in an AI class actually different between the two cities? Or is it the same thing with different weather?"

Good question, and the honest answer is: mostly the same thing, with a few real differences worth knowing before you pick a provider in either city.

## What's identical in both markets

Start here, because it's the part that matters most and the part most comparison posts skip.

- **The format that works is build-based, not theory-based**, in both cities. A child should leave with a real, deployed web app at a live URL — not a certificate, not a worksheet. This holds regardless of city.
- **The age sweet spot is the same**: 10-11 is workshop territory with a parent nearby, 12-14 is the strongest cohort for self-directed building, 15-17 can run at adult pace. See [what age should a child start AI classes](/blog/what-age-should-a-child-start-ai-classes) for the full breakdown.
- **The red flags are identical**: "AI awareness" with no building, class sizes above 15-20, worksheets and multiple-choice "AI safety" modules standing in for actual AI work, no clear answer to "what does my child walk out with?"
- **Real tools, not kid versions, matter in both cities.** Claude, ChatGPT, Figma, Midjourney — the professional tools, with adult supervision rather than artificial guardrails.

If you take one thing from this post: judge a provider in either city by the same test — will my child have a live URL to show me by the end?

## Where the two markets actually differ

### 1. Market maturity

Singapore's AI-for-kids market went from roughly zero to fifteen-plus visible providers in about six months, which means more noise but also more genuine choice. Hong Kong's market is smaller and leans heavily on an already-deep STEM-tutoring and robotics sector, a lot of which has simply relabelled existing curriculum as "AI." Practically: expect to filter harder in Hong Kong for providers that are actually teaching AI-building rather than robotics with new branding.

### 2. Pricing (once converted, it's closer than the raw numbers suggest)

- **Online weekly cohort** — Singapore SGD 60–100/session · Hong Kong HKD 350–600/session
- **In-person workshop** — Singapore SGD 150–200/session · Hong Kong HKD 900–1,400/session
- **Holiday camp (full week)** — Singapore SGD 600–1,200 · Hong Kong HKD 3,500–7,000

The HKD numbers look much bigger, but once you convert (roughly HKD 6 ≈ SGD 1), the real cost bands land close together. Full breakdowns with what you're paying for at each tier: [Singapore price guide](/blog/ai-classes-cost-singapore-price-guide) and [Hong Kong price guide](/blog/ai-classes-cost-hong-kong-price-guide).

### 3. Language and venue culture

Both cities teach in English by default for this kind of programme, which matters since Claude Code and most AI tools work natively in English. Hong Kong providers should confirm English-medium delivery explicitly if you're not already sure — most international-school and bilingual families are fine either way, but it's worth a direct question rather than an assumption. Venue-wise, Singapore's CBD-adjacent options (JustCo Marina Square and similar) map to Hong Kong's Central/Causeway Bay/Kowloon equivalents — both cities have compact, MRT/MTR-accessible options if you're choosing in-person.

### 4. School appetite

Both markets have international schools moving from "ban it" to "teach it," but Singapore's schools sector has moved slightly faster on whole-school AI hackathon days, partly because the after-school enrichment culture here was already dense before AI showed up. Hong Kong is close behind. If you're advocating for this at your child's school in either city, see [why international schools should teach AI](/blog/why-international-schools-should-teach-ai).

## If your family splits time between both cities

This is worth a specific note, because it comes up more than you'd think for Singapore/Hong Kong families. The cleanest path is a provider that runs the *same* curriculum, tools, and instructor standard in both cities — so your child isn't restarting from a different baseline each time you relocate for a term. Ask directly: "is this the same programme in both cities, or a different local partner?" The answer changes what continuity you can expect.

## Common questions

**Is AI education for kids different in Singapore vs Hong Kong?**
The core teaching approach is identical — build-based, small cohorts, real tools, a shipped project every time. What differs is market maturity (Singapore's market is more crowded and more mature), pricing (Hong Kong runs roughly 5-6x the SGD numbers in HKD, which is a similar real cost once converted), and venue culture (Hong Kong's STEM-tutoring sector is deeper, so more providers are relabelled robotics camps).

**Which city has better AI classes for kids, Singapore or Hong Kong?**
Neither is objectively better — the same red flags and green flags apply in both. The difference is how hard you have to search: Singapore has more options to filter through, Hong Kong has fewer but a higher share of relabelled STEM/robotics providers to watch for.

**Can a family with kids in both Singapore and Hong Kong use the same AI programme?**
Yes, if the provider runs in both cities with the same curriculum and instructors. That consistency matters for families that split time between the two — same tools, same format, same standard, regardless of which city the child is building in that term.

**Is online AI learning a good substitute for in-person in either city?**
For under-14s, in-person wins in both cities — the format benefits from an instructor who can lean over and look at a child's screen. Over 14, online works well in either market, since self-direction is the bigger factor than city.

## Pathwise across both cities — at a glance



## Where to go from here

We run the same **Idea to App** programme, same instructors' standard, same build-based format, in both cities:

- [Singapore programme](/singapore) — in-person at JustCo Marina Square, online across the region
- [Hong Kong programme](/hong-kong) — in-person at HK venue partners, online across the region

For the city-specific detail — pricing, venues, what to look for — read the full [Singapore guide](/blog/ai-classes-for-kids-singapore) or [Hong Kong guide](/blog/ai-classes-for-kids-hong-kong). Whichever city, the test that matters is the same: does your child walk out with a real, live URL.

— *Mr. Brown*


---

## Is Claude safe for my 10-year-old? A teacher's honest take

> An honest, non-corporate answer to the question parents actually ask. What 'safe' means in three dimensions, what Claude does well, what to watch for, and how to set up an account for a child the right way.

Published: 2026-06-07 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/is-claude-safe-for-my-10-year-old

**Short answer: yes, with three caveats.** Claude is one of the safer general-purpose AI tools for a 10-year-old in 2026 — Anthropic has invested unusually heavily in safety training, and the real-world failure modes are well-known and manageable. The three things to actually watch for are: **(1) the cognitive-offload trap** (the kid stops thinking and just asks Claude), **(2) the dependency loop** (Claude becomes the first-choice tool for everything), and **(3) the privacy hygiene gap** (they share things they shouldn't because the chat *feels* private). The fixes are all light-touch: shared account, weekly check-ins, a few ground rules about what not to paste. None of it requires lockdown.


A parent emailed me last month, two days before our school holiday camp:

> "I want my child to come. But I'm nervous about Claude. Is it safe? Like, *really* safe? Or is that something you say to get bookings?"

It's the right question. Here's the honest answer I sent back — written out properly, with no corporate softening.

## What "safe" actually means here

When parents ask "is Claude safe for my kid," they're usually asking three different questions wearing one coat. Worth pulling them apart.

### Dimension 1 — Content safety

*Will Claude show my child something it shouldn't?* Things like graphic violence, sexual content, instructions for harm, harassment, hate speech.

### Dimension 2 — Privacy and data

*What does Anthropic do with what my child types?* What happens to the chats? Is the child being profiled? Could a stranger end up with their conversations?

### Dimension 3 — Cognitive dependency

*Will my child stop thinking for themselves?* The "AI does my homework" trap. The "I'll just ask Claude" reflex. The atrophy of the muscle you actually wanted them to build.

These are three different problems with three different answers. Let's go.

## What Claude does well by default

Of the major frontier AI models in 2026, Claude is the one Anthropic has invested most heavily in training for honesty, harmlessness, and refusing genuinely dangerous requests. That's not marketing — it's a documented pattern, visible across years of model releases.

In practice, what that means for a 10-year-old:

- **Content safety is genuinely strong.** Claude refuses graphic violence, sexual content, and instructions for self-harm without me having to add a single filter. After running classes with hundreds of young builders, I've yet to see a kid stumble into something genuinely unsafe by accident. The refusals are clear and kid-readable — not lawyer-speak.
- **The default tone is good.** Claude doesn't talk down to kids, doesn't get sycophantic, doesn't push children toward parasocial attachment the way some chat-companion apps do. It will tell a kid when their idea won't work, kindly, and explain why. That's actually a great teaching pattern.
- **Hallucination on factual questions is well-managed.** Claude is more willing than most models to say "I'm not sure" or "you should check this against a source" — which is exactly the modelling you want for an 11-year-old learning to use AI as a thinking tool rather than an oracle.

If you only read one sentence in this section: **Claude is, as of 2026, the AI tool I'd most readily put in front of my own students.** That's not a paid endorsement. It's why we standardised on it.

## What to actually watch for

The honest failure modes — none of which Anthropic can fully solve for you, because they're behavioural rather than technical.

★ Failure mode 1 — Cognitive offload

The biggest real risk isn't Claude saying something wrong. It's Claude saying something *right*, every time, so quickly that your child stops doing the thinking themselves. The classic version: "I don't know, I'll just ask Claude." This is the same trap calculators created for arithmetic and Google created for trivia — just compressed into a year.

**The fix is behavioural, not technical.** Three small rules I'd recommend:
- For schoolwork, Claude is allowed to *explain* and *check* but not to *do*.
- The first answer to any question they give back to me has to be in their own words, not Claude's.
- Once a week, ask them: *"What's something you figured out without Claude this week?"*

★ Failure mode 2 — Dependency loop

The kid starts opening Claude before they've thought for ten seconds on their own. Decisions, opinions, what to read next, what to draw. Claude becomes the default scaffolding for everything.

This is less about "Claude is bad" and more about "ten-year-olds will lean on any nearby crutch." The fix is the same fix that worked for early-2000s parents and "Google it" — model the behaviour of *thinking first, asking second*. Visibly. Out loud.

★ Failure mode 3 — Privacy hygiene

A chat with Claude *feels* private — like a journal or a chat with a friend. It isn't. Anthropic processes the conversation. Other people (parents, teachers, in some cases anonymised model training) may end up seeing samples down the line.

What this means in practice: **kids should not paste into Claude anything they wouldn't write on a postcard.** Their home address, their school's name and class, their friends' real names + situations, their parents' financial details. Most kids will do all of these things at least once before they get the rule. The fix is to say it out loud, then say it again in two months.

## How to set up an account for a child (the right way)

The setup that's worked best across the hundreds of young builders I've taught:

1. **Account in a parent's name, shared with the child.** Anthropic's terms require accounts to be 18+, so the account belongs to you and the child uses it with you. Practically, this also gives you visibility into chat history without surveillance theatre — they know you can see, and the conversation just happens differently when both of you can.
2. **Pay tier, not free tier, if you can.** Pay tier conversations get stronger privacy treatment (Anthropic does not train on Pro conversations by default). It's USD ~20/month. For a kid who uses it twice a week, this is the single highest-leverage US$20 you'll spend.
3. **First three sessions, you're in the room.** Not hovering. Just present. You're watching for how they *use* it more than what they ask. That's where the habits get set.
4. **One "house rule" sticker by the laptop.** Some version of: *"Think first. Ask Claude second. Never paste anything you wouldn't write on a postcard."* You'll be surprised how much that prompts before they hit return.
5. **Weekly 3-minute check-in.** Three questions: *"What did Claude help you with this week? What did you figure out without it? Did anything weird happen?"* Three minutes. Once a week. That's the whole parental routine.

## When to back off

Most parents I talk to over-engineer the supervision and then quietly drift to under-engineering it once the kid is "fine." The reverse is healthier.

- **Weeks 1–4**: in the room for first sessions, weekly check-in, strict house rule.
- **Months 2–6**: occasional check-in, ask to see one chat a fortnight, trust building.
- **Beyond 6 months**: weekly question, intervene only if you see a specific failure mode show up (homework copy-paste, weird sharing, dependency creep).

The goal is a kid who can think with AI, not a kid who needs you sitting next to them to think with AI. The supervision should slowly retire itself.

## A note on AI safety classes for parents

If you want to go deeper than this post, the [step-by-step guide to introducing AI to your 10-17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old) covers the parent-facing setup in more detail. The [what to look for in AI classes for kids](/blog/ai-classes-for-kids-singapore) post is the Singapore equivalent for paid-class quality, and the [Hong Kong version is here](/blog/ai-classes-for-kids-hong-kong).

If your child is going to use AI either way — through school, through curiosity, through a friend's recommendation — the better question isn't *"how do I keep AI away?"* It's *"how do I make sure the first 50 hours they spend with it are good ones?"* That's the bet our [young builder programs](/young-builders) are built on: structured first hours, real builds, real teacher in the room, real conversations about exactly the failure modes above.

If you want to ask anything specific — a setup question, a "but what about *this* scenario" — [drop us a line](/contact). I read every parent message myself.

The hardest part of AI safety for a 10-year-old isn't the model. It's the routine around it.

— *Mr. Brown*


---

## I shipped an iPhone app entirely with AI in 30 days — what actually worked

> A teacher's first-person account of building and shipping a real iPhone app to the App Store using Claude Code, with no prior iOS experience. What worked, what didn't, and what it means for what we teach.

Published: 2026-06-04 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/30-day-iphone-app-with-ai

The moment I knew the App Store would actually let it in, I was sitting on my kitchen floor at 11pm, refreshing App Store Connect for the eighth time. The email finally landed: *Your app is now available on the App Store.*

I shouted. My wife asked what happened. I tried to explain. It came out as: *I made an iPhone app. I do not know how to make iPhone apps.*

A few weeks later, a 12-year-old in one of my Saturday cohorts asked the same thing: "Wait — you *made* this?" And I realised the most valuable thing wasn't the app. It was the receipts: a real, shipped, on-the-App-Store project built by someone who, four weeks earlier, had never opened Xcode.

Here's what actually happened.

## The app

[PLACEHOLDER: One-sentence description of the app — what it is and who it's for. Example: "A small daily-habit tracker called PocketCue that nudges you with one question each morning."]

I built it because [PLACEHOLDER: short reason — e.g. "I wanted something simpler than the habit apps I'd tried" / "my own kids needed it" / "I wanted a real artifact to show my adult students"].

You can find it on the App Store here: [PLACEHOLDER: App Store URL once published].

## My starting point

I want to set the baseline honestly, because it matters for whether this story is useful to anyone else:

- **Last time I wrote production code**: many years ago, in a different language, in a different life.
- **iOS experience**: zero. I'd never opened Xcode. I didn't know what SwiftUI was. I genuinely thought "TestFlight" was a feature on actual aeroplanes.
- **Daily tools at the start**: a teaching laptop, an Anthropic account, a hand-me-down iPhone, and a fresh Apple Developer Program subscription (USD 99/year, the only real upfront cost).

If you can teach, write a clear email, and read English, your starting baseline is close to mine.

## The stack

The whole thing ran on a small, surprisingly modern stack:

- **[Claude Code](/blog/what-is-claude-code)** as the primary builder. This was the engine. It wrote the Swift, debugged the Swift, ran the Swift, and made me feel like the architect rather than the labourer.
- **Xcode** as the harness Claude Code drove. I opened it almost only to hit ⌘R.
- **SwiftUI** as the UI framework. I learned what it was in week one, by reading what Claude Code had written for me.
- **[PLACEHOLDER: Backend or storage choice — e.g. "Supabase for auth + sync" / "SwiftData for local-only storage" / "No backend; everything's on-device"]**.
- **TestFlight** for beta testing.
- **App Store Connect** for the submission gauntlet.

Total monthly cost of the stack while building: roughly USD 20 (Anthropic Pro) + the one-time Apple developer fee.

## The 30-day timeline

It was actually more like 28 days end-to-end, but "30 days" reads better. Here's how it broke down.

### Week 1 — From idea to "wait, this works on a real phone"

The first day was a single prompt to Claude Code: *"Build me a SwiftUI iOS app that does X. Pick the simplest possible architecture. Use only Apple frameworks unless we have to."* It wrote a scaffold. Xcode complained about signing. Claude Code walked me through the signing dance. By the end of day two, I had a "Hello, world" version running on my own phone via cable.

By the end of week one, the core flow worked. Ugly, but worked. I texted screenshots to one friend who said "you made *that*?" in a tone that I think was meant as a compliment.

### Week 2 — Real features

This is where the slope steepened. Adding the third feature broke the second. Adding the fourth broke both. Claude Code was unfazed; I was constantly. The lesson of this week was *commit small, run often.* Every working state got a git commit. Every bug got described to Claude Code in full sentences instead of one-word panic.

[PLACEHOLDER: One specific feature I'm proud of from this week — e.g. "the streak-counter animation" / "the notification-permission flow that doesn't feel scammy"].

### Week 3 — Design and polish

The week I learned that "shipped" and "good" are different things. The app worked. The app was ugly. Claude Code helped me build a coherent visual system — typography, spacing, colours, a single accent — by describing the *feel* I wanted ("calmer than Notion, warmer than Apple's defaults"). It produced screens I'd never have arrived at on my own.

This week was also where I got the App Store screenshots, the icon, the privacy policy text, and the support page set up. All boring. All necessary. Claude Code drafted most of them; I edited.

### Week 4 — Ship

Submitted to App Store review on day 24. First rejection on day 25 ([PLACEHOLDER: brief reason — e.g. "they wanted a clearer privacy disclosure" / "they wanted me to add a 'delete account' option"]). Fixed it the same evening. Resubmitted. Approved on day 28. Live on day 28.

## 5 things that actually worked

The honest list of patterns I'd recommend to anyone doing this:

1. **Commit after every working change.** This sounds like engineering hygiene, but it's actually survival. When Claude Code makes a change that breaks three things, the answer "go back to the last commit and try again" is the difference between calm and panic. Make this your only rule.
2. **Describe in full sentences, not keywords.** "Bug" is useless. "When I tap the streak button on day two, the count resets to one instead of two" is gold. The quality of your prompt is the quality of your fix.
3. **Let it pick the boring stuff.** Architecture, file structure, naming, error handling, accessibility scaffolding. Claude Code's defaults here are usually good. Don't override unless you know why.
4. **Run on a real phone from day one.** The simulator lies about things — gestures, scroll feel, keyboard behaviour. Cable, real device, every day.
5. **Treat App Store review as part of the build.** Read the App Store Review Guidelines before you submit. Most rejections are predictable. Mine was.

## 3 things I'd do differently next time

The honest list of things that cost me time:

1. **I designed too late.** I treated design as a "polish week" rather than a constraint from day one. Result: I rebuilt several screens because they didn't fit the visual system I eventually settled on. Next time, I'll pick the typography + colour + spacing system in week one, not week three.
2. **I overbuilt the data model.** I planned for features I never shipped. Claude Code happily produced complex schemas that turned out to be dead weight. Next time, I'll resist the urge to "make it future-proof" and just build for what's in the next two weeks.
3. **I didn't have a real beta tester until week three.** TestFlight was sitting right there. Next time, I'll get one person who isn't me using the build by day five. The bugs they find in twenty minutes are the bugs that would have eaten my whole week three.

## What this means for what we teach

The reason I'm writing this isn't bragging rights. It's that the experience changed how I teach.

Before this app, I'd tell adult cohorts and ambitious teens that *yes, you can build real things with AI* — but I was teaching the abstract version. Now I can teach the concrete version. I've felt the panic of a broken build at 2am. I've debugged a SwiftUI ChartView I didn't understand. I've stared at an App Store rejection email and figured out what they actually wanted. Every lesson I teach is now grounded in something I've actually done.

If you're an [adult learner](/adults) wondering whether vibe coding can take you past "toy projects" — yes, it can. If you're a [parent of a teenager](/young-builders) wondering whether the skill compounds — yes, it does. The same Claude Code that built my iPhone app is the tool every cohort uses to ship their first web app.

If you want to see the [Pathwise programs that teach this](/programs/vibe-code), or read the [vibe coding primer](/blog/what-is-vibe-coding) that explains the philosophy, or just want the [next cohort dates](/classes) — they're all one click away.

The hardest part of shipping an iPhone app with AI wasn't the AI. It was deciding the app deserved to exist.

— *Mr. Brown*

---

*PS — placeholders in this post (`[PLACEHOLDER: …]`) are the spots only the founder can fill: actual app name + description, the specific stack choices, one concrete pride-of-week-two feature, the real rejection reason. Replace before publish, or remove these notes if the post is being kept as a generic-but-honest founder reflection.*


---

## AI classes vs bootcamps vs online courses: the best way to learn AI in 2026

> A teacher's honest, balanced comparison of self-paced online courses, multi-week coding bootcamps, and short build-based AI cohort classes — strengths, costs, and who each suits in 2026.

Published: 2026-06-03 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-vs-bootcamps-vs-online-courses

There's no single best way to learn AI in 2026 — it depends on your goal. Self-paced online courses are cheap and flexible, but most people never finish them. Coding bootcamps build real depth over weeks or months and suit career-changers aiming at engineering jobs. Short, build-based cohort classes get you to a real, deployed app fastest, which is what most people actually want. The honest test for any of them: do you finish with something real on the internet, or just a certificate?


I get a version of this question every week, from adults and parents alike:

> "I want to learn AI. Do I do one of those online courses, a proper bootcamp, or one of your classes? What's actually the best way in 2026?"

It's a fair question, and I'll give you a fair answer — not a sales pitch dressed as advice.

All three formats are legitimate. They're built for different people with different goals, time budgets, and definitions of "done." Pick the wrong one for your situation and you'll either waste money or quit before the payoff.

Here's how I'd think it through.

## First, get honest about your goal

Before comparing formats, answer one question: *what do you actually want at the end?*

- A *job* as a software engineer? That's a real, specific target — and a demanding one.
- *Background knowledge* so you stop feeling lost in meetings? Reasonable, and cheap to get.
- A *real thing you built* — a working app, a tool that saves you hours, a prototype you can show? This is what most people want when they say "learn AI," even if they don't phrase it that way.

The right format falls out of that answer almost automatically. Most of the regret I see comes from people who picked a format for a goal they didn't actually have.

## Self-paced online courses

The Udemy / Coursera / YouTube end of the market. You buy — or stream for free — a library of recorded videos and work through them whenever you like.

The strengths are real. They're cheap, often under the price of a dinner. They're flexible: 6am, midnight, on the train, whenever. And the good ones are genuinely well-produced, taught by people who know their material.

The weaknesses are just as real. Completion rates are notoriously low — most people who buy a course never finish it, and I'd bet you already have one or two unfinished ones in an account somewhere. There's no one to ask *"wait, why did it do that?"* the moment you're stuck, which is exactly when most beginners give up. And because the content is recorded, it ages fast in a field where the tools change every few weeks.

The deeper issue: watching is not doing. You can finish ten hours of video about AI and still have built *nothing*. The knowledge feels real until you sit down to make something and realise none of it transferred into your hands.

**Who it suits:** disciplined self-starters who want background knowledge, people on a tight budget, or anyone topping up a specific skill they already half-have. A fine first toe in the water. A poor place to actually become capable.

## Coding bootcamps

The intensive end. Multi-week and often multi-month programs — full-time or part-time — that aim to take you from beginner to employable developer. Traditionally they're syntax-first: you learn to write real code, build projects, and grind through fundamentals.

The strengths are depth and structure. A good bootcamp builds durable fundamentals — how software actually works under the hood — that no shortcut gives you. There's a cohort, a schedule, accountability, and often career support at the end. For someone genuinely changing careers into engineering, that depth is the point.

The weaknesses are the size of the commitment. It's frequently the most expensive option by a wide margin, and it costs weeks-to-months of your life. The intensity is real, and so is the burnout. And here's what's changed: a lot of traditional bootcamp curriculum was designed for a pre-AI world, where writing every line by hand was the only way to ship. In 2026 that's no longer true — and not every bootcamp has caught up.

**Who it suits:** people aiming squarely at a professional software-engineering role, who have the months and the money, and who *want* the deep fundamentals. If that's you, a bootcamp can be genuinely transformative. If it's not, it's a lot of machinery for a job you didn't need done.

## Short, build-based AI cohort classes

This is what we run, so read the next part with appropriate suspicion — I'll keep it honest, limits included.

These are short, small-group classes where you use AI tools to build and *deploy* real things, fast. You don't grind syntax for weeks before you make anything. You describe what you want, direct the AI to produce it, learn to read and fix the result, and ship it live — often in your very first session.

The strengths are speed to a real, working result, and live instruction — so the moment you're stuck, there's someone looking at *your* screen. Cohorts are small, which means actual attention per person. And you leave with a deliverable that proves the learning landed: a live URL you can show anyone.

The honest weaknesses: a short class will not, on its own, make you a professional software engineer — that's not what it's for, and any provider who claims otherwise is overselling. It's less deep than months of fundamentals. And because the format depends on small groups and live teaching, it costs more than a recorded video course, though far less than a full bootcamp.

**Who it suits:** adults who want to *build* rather than *talk about* AI — founders who want to prototype, operators who want to automate a workflow, curious people who want to feel competent fast. And parents choosing for a 10-to-17-year-old, where shipping something real in week one is what sustains a kid's attention in a way syntax drills can't. For more on the younger end, see our take on [AI classes vs coding classes for kids](/blog/ai-classes-vs-coding-classes-for-kids).

## The side-by-side



A note on cost, because everyone asks: I won't quote our class prices here — they vary by format and are shown when you book. But for honest market context on what hands-on AI classes actually run in this region, I've written full breakdowns for [Singapore](/blog/ai-classes-cost-singapore-price-guide) and [Hong Kong](/blog/ai-classes-cost-hong-kong-price-guide).

## The one test that cuts through all of it

If you remember nothing else, remember this question — and ask it of *every* option:

> "At the end of this, do I have something real — on the internet, that I made — or do I have a certificate?"


**The "something real" checklist.** Before you pay for any AI learning format, get a clear answer to:

- What, specifically, will I have built and deployed by the end?
- Is there a *live* result — a URL, a working tool — or just notes and a quiz?
- Can I ask a real human "why did it do that?" in the moment I'm stuck?
- Are the tools the actual ones professionals use, or a sandbox?
- How many people share one instructor's attention?

If a provider can't answer the first one plainly, that's your answer.


A certificate proves you attended. A deployed app proves you can do the thing. In 2026, with AI doing the heavy lifting on the code, the portfolio *is* the qualification — and "I built and shipped this" beats "I completed a course on this" in almost every room you'll walk into.

## So which should *you* pick?

Let me make it concrete.

- **You want cheap background knowledge and you're disciplined:** start with a free or low-cost online course. Just be honest with yourself about whether you'll finish it.
- **You're changing careers into professional software engineering, with months to spend:** a bootcamp's depth is worth it. Check that its curriculum has caught up to how software is actually built now.
- **You want to build and ship real things with AI — soon, without quitting your job:** a short, hands-on cohort class is the fastest route to a working result you keep.
- **You're choosing for a 10-to-17-year-old:** lead with building. The shipped result is what keeps a kid hooked long enough to *want* the deeper stuff later.

And these aren't mutually exclusive. The order I'd actually recommend for most adults: build something real first to confirm you love it, *then* go deeper if you're hooked — not the other way around. Motivation is the scarce resource, and nothing builds it like seeing your own thing live on the internet.

## Where to go from here

If you've decided you want to *build* — not just watch videos about building — that's exactly what we do. We run [hands-on cohort classes for adults](/adults) across [Singapore](/singapore), [Hong Kong](/hong-kong), and online for anyone in APAC, and every learner walks out with a real, deployed app and a certificate to go with it — not instead of it. Have a look at the [upcoming schedule](/adults#schedule) and pick a date that fits.

Whatever you choose — course, bootcamp, or cohort — pick the one that ends with you having *made* something. That's the version of "learning AI" that actually changes what you can do on Monday.

— *Mr. Brown*


---

## How to pitch an AI hackathon to your school leadership (and get buy-in for AI in the classroom)

> A practical guide to getting buy-in for AI in the classroom: how to frame the case, answer the cost, safety, and disruption objections, and propose a low-risk hackathon pilot your SLT will approve.

Published: 2026-06-03 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/how-to-pitch-ai-hackathon-to-school-leadership

The easiest way to get buy-in for AI at your school is to pitch a small, contained pilot — usually a single-year-group AI hackathon — not a whole-school overhaul. Frame it around outcomes leadership already wants (engagement, real student work, parent stories), bring the answers to the cost, safety, and disruption questions *before* they're asked, and make it a clean one-day "yes" with a fixed start and end. Send your SLT the [anatomy of a hackathon day](/blog/how-we-run-an-ai-hackathon-at-a-school) so they can picture it, then offer a [free discovery call](/schools#book-discovery) so they can ask their own questions.


Most of the teachers who email me about running an AI event at their school have already done the hard part. They *get it.* They've seen what a student can build in an afternoon. They want it for their kids.

What they don't have is a way through their own leadership.

"My head of school is interested but cautious." "The board will ask about cost." "Our safeguarding lead will have questions." "I don't want to look like I'm chasing a trend."

I've sat on the other side of these conversations many times now. So here's something genuinely useful — not a pep talk, but the actual playbook for getting AI in the classroom approved by people whose job is to be careful.

## Start with the right ask (this is where most pitches fail)

The single most common mistake is asking for too much.

A teacher walks into the SLT meeting and says *"we should embed AI across the curriculum."* Leadership hears: cost, risk, training, policy, parent emails, and a multi-year commitment they can't evaluate. So they say *"let's revisit this next year."* And it dies.

Don't ask for the cathedral. Ask for one brick.

The lowest-risk, highest-signal version of "let's try AI" is a **single-year-group hackathon day.** One cohort. One day. A fixed start and end. It touches nothing else in the school. If it goes brilliantly, you have evidence and momentum. If it's just *fine*, nothing structural has changed and nobody's reputation is on the line.

That asymmetry is your whole pitch. A pilot is a small, reversible bet with a visible upside — and small reversible bets are exactly what cautious leadership is built to approve.

## Frame the case in their language, not yours

You are excited about the technology. Your principal is accountable for outcomes. Translate.

Don't pitch "AI." Pitch the things your leadership *already* loses sleep over:

- **Engagement.** A hackathon is the most focused energy you'll see in the building all year. Heads of school routinely walk in expecting chaos and find a quiet room of kids building.
- **Real student work.** Not a worksheet, not a poster — a live, deployed web app with a URL every student can show their parents. That's tangible in a way most "innovation" initiatives never are.
- **Admissions and parent comms.** The photos, the demos, the "my kid built something on the actual internet" stories — these go in newsletters, on Open Day, in the next admissions round. Leadership understands that currency immediately.
- **Future-readiness without the hand-waving.** Every parent in your community is already asking what the school is doing about AI. A real, hands-on event is a far better answer than a policy document.

When you frame it this way, you're not asking leadership to bet on a technology. You're offering them a win on metrics they already report against.

## Anticipate the five objections — and bring the answers first

Credibility comes from raising the hard questions *yourself*, before anyone else does. Here are the five that come up, and how to handle each.

### "What's this going to cost?"

Be honest that there's a cost, and contain it. A single-cohort pilot is a fraction of a whole-school programme, and you can frame it as a one-off trial spend rather than a line in next year's budget. Prices vary by format and cohort size and are shown at booking — but if leadership wants market context to sanity-check the number, the [Singapore](/blog/ai-classes-cost-singapore-price-guide) and [Hong Kong](/blog/ai-classes-cost-hong-kong-price-guide) price-guide posts give them a frame of reference before any conversation with us.

### "Is it safe for the students?"

This is the safeguarding question, and it's the right one to ask. The honest answer: we use AI tools with strong content guardrails and we monitor outputs throughout the day. Inappropriate content is essentially never an issue, and when an edge case comes up our mentors catch it and turn it into a teaching moment. Offer to put your safeguarding lead directly in touch with us — many schools want the approach documented ahead of time, and a provider who flinches at that request is telling you something.

### "Won't this disrupt the timetable?"

Run it on a day that's already off-timetable — an activities day, an end-of-term day, an enrichment block. A self-contained event doesn't pull teachers out of their teaching for weeks of prep. That removes the disruption objection almost entirely.

### "Our teachers don't know AI. Will they be embarrassed?"

No — because they aren't expected to lead anything technical. The mentors do that. Teachers do what they're already excellent at: pastoral care, reading the room, keeping things flowing. Most leave more curious than they came, having learned more by walking the room than they would in a formal training session. Make this explicit in your pitch, because "my staff aren't ready" is a quiet objection that kills more proposals than cost ever does.

### "Isn't this just teaching kids to cheat?"

The most important reframe. Cut-and-paste cheating is a student hiding AI use to avoid doing the work. A build day is the *opposite*: students openly directing AI to make something that didn't exist before — and being judged on their ideas, their judgement, and what they shipped. When leadership watches a student stand up and demo a working app they couldn't have made alone, the cheating worry tends to evaporate on its own.

## What evidence to bring

You don't need a research deck. You need three things that let leadership picture it concretely.


  **Bring to the meeting:**

  - **A picture of the day itself.** Send leadership the [hour-by-hour anatomy of a hackathon day](/blog/how-we-run-an-ai-hackathon-at-a-school) so "a hall full of students building at once" stops sounding like chaos and starts sounding like a plan.
  - **A clear split of who does what.** A short list of what the school provides (room, WiFi, laptops, a teacher liaison) versus what the provider brings (curriculum, mentors, tooling, deployment, safety approach). Ambiguity makes leadership nervous; a clean division makes it feel handled.
  - **A written proposal.** After a short discovery call, we send a written proposal a few days later. Walking into the SLT meeting with a document — not just enthusiasm — changes how you're heard.


The goal is to let leadership stop imagining the worst-case version in their head and evaluate the actual thing in front of them.

## What to put in the proposal

Keep your internal one-pager short. The more confined and specific it is, the easier it is to approve.



Notice what that last row does. By naming the *next* step, you quietly signal that the pilot is the beginning of a path — without forcing leadership to commit to the path today. You're giving them an easy "yes" now and an obvious "yes" later.

## Handle the logistics so leadership doesn't have to

The proposals that get approved are the ones where the teacher has clearly already thought about the boring parts. A clean pilot day takes roughly six to eight weeks from a signed brief to the event — enough time for theme scoping, mentor briefing, tooling provisioning, photo permissions, and a parent-comms package.

If you can walk into the meeting and say *"I've checked the activities-day calendar, the hall is free, I've spoken to facilities about pod seating, and the provider handles the AI accounts and deployment,"* you've removed nearly every reason to defer the decision.

Leadership says no to vague ideas and yes to things that look *organised.*

## Make the first conversation easy

Here's the move I'd actually recommend. Before you go to your SLT, [book a free discovery call](/schools#book-discovery) yourself. Bring us your school's priorities and constraints, and we'll help you shape the pitch and send a written proposal a few days later. Then you walk into leadership with a real plan, real answers, and a document — instead of asking permission to go figure it out.

Or send your principal straight to our [school programs page](/schools) so they can see the full menu — single-year-group events, whole-school days, multi-day intensives — and picture where a pilot could lead.

You've already done the hard part by caring enough to push. The rest is just making it easy for careful people to say yes.

The students will surprise your leadership. They always do.

— *Mr. Brown*


---

## What teams actually ship after a one-day AI workshop

> Corporate AI training that ends with something built, not a quiz. Here's what marketing, ops, sales, and leadership teams realistically ship after a one-day AI workshop — and how to judge the ROI.

Published: 2026-06-03 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/what-teams-ship-after-ai-workshop

After a one-day AI workshop, a team should walk out with a **real, working artefact tied to their actual job** — an internal tool, an automation, or a deployed prototype — not a certificate and a quiz score. In our corporate sessions, marketing teams typically ship a draft-and-review workflow, ops teams ship a small internal app or automation, and sales teams ship a research or follow-up assistant. The point of a hands-on build over slideware is simple: the team leaves able to repeat the process without us in the room. Formats run half-day to weekend, onsite or online, across Singapore, Hong Kong, and APAC.


An L&D lead asked me a fair question on a call last month: "If I book my team a one-day AI workshop, what do they actually *have* at 5pm? Be specific. I've sat through too many days that ended with a feedback form and nothing else."

So let me be specific.

This post is the honest version of what marketing, ops, sales, and leadership teams realistically *ship* after a single day with us — and how to think about whether it was worth it, without me waving made-up numbers at you.

## First, what "ship" means here

When I say a team ships something, I mean there's an artefact at the end of the day that exists outside the workshop. Something you can open on Monday. A URL. A working tool. A documented automation that runs.

Not notes. Not a deck. Not "exposure to concepts."

That distinction is the whole game. A team that *watched* a day of AI content has a vocabulary. A team that *built* for a day has a skill they can repeat — and, crucially, an artefact their colleagues can see and ask about. The thing on the screen is what turns one day into a habit.

So here's what that artefact tends to look like, by team.

## What marketing teams ship

Marketing is usually the easiest team to point at a quick win, because so much of the work is repeatable production with a human-judgement layer on top.

The common build is a **draft-and-review workflow** for something the team does every week — a campaign recap, a batch of social posts in the brand voice, a first-draft newsletter, a competitor scan. The team brings their actual brand guidelines and a real past example, and we build a workflow that produces a *first* draft in their tone, with the human firmly in the review seat.

The important word is *draft*. Nobody walks out having automated their judgement away. They walk out having removed the blank-page friction between an idea and a usable starting point.

## What ops teams ship

Ops is where I most often see a small **internal tool** get built — the kind of thing that used to require a developer ticket and a three-week wait.

A simple intake form that sorts and routes requests. A checklist app that enforces a process everyone keeps skipping. A little dashboard that pulls a messy spreadsheet into something readable. A document that drafts itself from a few inputs.

These are unglamorous, and that's exactly why they matter. They're the small recurring annoyances that never make it onto a roadmap because they're not big enough to justify engineering time — but they quietly cost the team an hour here and an hour there, every week.

## What sales teams ship

Sales teams usually leave with an **assistant for the parts of the job that aren't selling.**

Pre-call research pulled together from a few sources. A follow-up draft that's specific to the conversation instead of a template. A way to turn rough call notes into a clean CRM entry. A first pass at a proposal from a short brief.

Same principle as marketing: the human stays in the relationship. The build just removes the admin tax that eats into the hours reps would rather spend talking to people.

## What leadership teams ship

Leadership sessions are a little different. Sometimes a leader builds a tool of their own — a board-update drafter, a quick way to summarise a long thread. But often the more valuable thing they ship is *judgement*: a grounded, first-hand sense of what this technology can and can't do for their org, built by actually doing it rather than reading a forecast about it.

A leader who has spent a day building is a much harder person to sell vapourware to. That alone has paid for a few of these workshops.



## Why a build beats slideware

I'll keep this short, because I've made the longer case in [what real AI training for teams looks like](/blog/what-real-ai-training-for-teams-looks-like). But the core of it is three things.

**Watching is not doing.** Nobody learns to ride a bike from a deck about bikes. AI is the same. The understanding only arrives once you've sat with a real tool, given it real input, gotten something wrong, and fixed it.

**Generic examples don't transfer.** "Here's how AI *could* help with sales emails" lands as theory if the email isn't yours. The bridge from the example to the actual job gets left unbuilt, so people nod and then never act on it. We build on your work, with your data, so there's no bridge to cross.

**No artefact, no momentum.** A team that leaves with a certificate has nothing to point at next Monday. A team that leaves with a working tool has a thing their colleagues notice — and noticing is how the second build, and the third, get started.

## How to think about ROI (without me inventing a number)

People ask me for the ROI figure. I'd rather not make one up, and you should be suspicious of anyone who does.

The honest framing isn't *"how many dollars per training hour did we save?"* It's *"how much faster can the team move on things they already want to do?"* Good AI training doesn't replace anyone's job — it removes the friction between an idea and the output. The unlocked velocity is the return. It's real, and it's genuinely hard to put a clean number on in a quarterly review.

What I *can* tell you is how to test it yourself.


**The six-week test.** A month and a half after the workshop, ask the team one question: *"What's something you do differently now?"* If a few people can name something specific without pausing — a tool they still use, a task that takes half as long — the day worked. If the answer is "we talk about AI more, I think?" it didn't. That's a far more trustworthy signal than any ROI slide.


The teams that get the most out of a workshop usually do one small thing afterward: they pick *one* of the things they built, give it an owner, and agree to actually use it for two weeks. That's it. The build creates the possibility; a named owner turns it into a habit.

## Choosing a format

A rough guide, since this is the question that follows next:

- **Half-day** is a focused single build. Best for one team with one clear, shared pain point.
- **Full-day** lets a few small groups each build a different thing, then demo to the room. Best when a team has several recurring annoyances rather than one big one.
- **Weekend intensive** goes deeper — bigger prototypes, more iteration, more "what do we roll out across the company?" conversation. Best for leadership or a team treating this as a genuine kickoff.

We pre-scope every one of these with the team lead beforehand, so the room walks in pointed at a real outcome instead of "AI literacy in general." Prices vary by format and team size and are shown at booking — if you want market context on how this kind of training is priced in the region, the [Singapore](/blog/ai-classes-cost-singapore-price-guide) and [Hong Kong](/blog/ai-classes-cost-hong-kong-price-guide) price guides are a fair starting point.

## So — what will *your* team ship?

I genuinely don't know yet, and that's the point. The answer depends on the recurring annoyance your team would most like to hand off, and we figure that out together before the day.

If you've got one in mind — the Monday report nobody wants to write, the intake process held together with copy-paste, the follow-ups that pile up — that's exactly the raw material for a workshop. Have a look at how we run these on the [companies](/companies) page, then book a free discovery call and tell me what your team would build. No slides on that call either.

— *Mr. Brown*


---

## Why international schools in Singapore & Hong Kong should teach AI in 2026

> The future-skills case for AI programmes in international schools — why banning it backfires, what good build-based AI in schools looks like, and how it fits your existing curriculum.

Published: 2026-06-03 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/why-international-schools-should-teach-ai

International schools in Singapore and Hong Kong should teach AI in 2026 — not ban it. The skill that matters is directing AI to build and create, then judging the result, and it's becoming a baseline literacy the way typing or research once was. The version that works is *build-based* — students ship real, deployed apps — not lecture-based or ban-based. Done well, it strengthens your existing curriculum rather than competing with it. The fastest way to scope it for your school is a [free discovery call](/schools).


I have this conversation with heads of school more than any other, and it usually starts the same way.

"We know we need to do *something* about AI. We're just not sure what. Half our staff want to ban it, half want to embrace it, and nobody's quite sure what 'embrace it' even means in a classroom."

That's an honest place to start. So let me give you the honest version of the answer — the case for teaching AI in your school in 2026, what "good" actually looks like, and the worries that come up every single time.

## The ban doesn't work — and it's the riskiest option

The instinct to ban is understandable. AI feels like a threat to academic integrity, and a blanket policy feels like control.

But here's what a ban actually does: it pushes AI use into the dark. Students don't stop using it. They use it at home, on their phones, on their own accounts, with no one teaching them when it's honest, when it's lazy, when the output is wrong, and when they're being quietly misled by a confident-sounding paragraph.

A ban doesn't remove AI from your students' lives. It just removes *you* from the room where they learn to use it.

That's the part I'd want every head of school to sit with. The students who most need an adult guiding their AI use are the ones a ban leaves most alone with it.

## The future-skills case, stated plainly

I try to avoid the breathless "AI changes everything" talk — it's mostly hype, and you've heard enough of it. So let me make the narrow, defensible version of the argument.

The ability to direct an AI tool — to describe what you want, judge what comes back, spot when it's wrong, and refine until it's right — is becoming a baseline skill across most knowledge work. Not a specialist skill. A *baseline* one, the way using a search engine or a spreadsheet became baseline.

Your students will graduate into a world where this is assumed. The question for a school isn't whether they'll use AI. It's whether they'll arrive with judgement or without it.

And judgement is the word that matters. You don't get judgement from a policy document or a one-off assembly. You get it from doing — from making things with AI, getting it wrong, fixing it, and slowly developing taste for what good looks like.

## What "good" actually looks like: build, don't lecture

Here's where most school AI efforts go sideways. They treat AI as a *topic* — a unit, a slide deck, a discussion about ethics and the future of work.

Those discussions have their place. But a slide deck about AI builds exactly as much real skill as a slide deck about swimming.

The version that works is build-based. Students make a real thing — and at Pathwise that thing is always a live, deployed web app with a URL they can send to their family. Not a worksheet. Not a poster. Something that works, on the internet, that they made.

When a twelve-year-old deploys their first working app and realises a grandparent overseas can open it on their phone *right now*, something shifts. The abstract becomes concrete. And concrete is where learning sticks.


**What separates a good school AI programme from a weak one:**

- Students *build and ship* something real — not just discuss AI
- Every learner leaves with a live URL, not a certificate of attendance
- Real professional tools, not a sandboxed toy version
- Small enough groups that an instructor sees every screen
- The teaching targets durable judgement, not this month's specific tool
- It maps onto your curriculum and priorities, not bolted on beside them


If a programme can't point to a real artefact at the end of it, you're paying for theory. Theory about AI is an article. It isn't a programme.

## The school offering, at a glance

When schools ask what working with us actually involves, this is the shape of it. We run on campus, across Singapore, Hong Kong, and the wider APAC region, and tailor the curriculum by age group.



The hackathon format is the one most schools start with, because it's the most visible — 200 to 500 students all building at once, every one of them shipping something by the end of the day.

## The worries — answered straight

In my experience, heads of school and curriculum leads ask versions of the same three questions. Let me take them head-on.

### "Is this just a fad we'll regret investing in?"

Fair. Schools have been sold "the future" before and ended up with a cupboard of interactive whiteboards nobody used.

The honest distinction is this: the *tools* change constantly — the specific app a student uses today may be replaced within a year. But the underlying skill — describing what you want, directing a tool to build it, judging the result — is stable. A good programme teaches that durable skill, not the tool of the month. So it doesn't go stale when the tools shift, because the tools shifting is exactly what we're teaching students to handle.

### "Is it safe?"

The version that worries you — a student typing something inappropriate into an unsupervised chatbot — is a real risk *of the unsupervised version*. It's much smaller in a properly run classroom.

We use tools with content guardrails, our instructors monitor outputs through the session, and we're happy to walk your safeguarding lead through our safe-usage approach before anything starts. Many schools want that documented, and we provide it. The point I keep coming back to: the *less* safe path is the one where students use consumer AI alone at home with no adult teaching them how.

### "Will it stop students learning to code?"

This is the one I hear from heads of computer science most, and I understand the fear. The honest answer is the opposite of what you'd expect.

Getting a student to a working app quickly is the single best thing for their appetite to learn fundamentals. Once they've shipped something real, they start asking *"but how does this actually work?"* — and that's the exact moment proper coding becomes worth teaching, because now they have a reason to care. Syntax-first, for most students, front-loads the boring part and loses them before the payoff. AI-first builds the motivation that makes the fundamentals land. It doesn't replace coding. It's the best on-ramp to it I've found.

## How it fits what you already do

The schools that get the most from this don't treat AI as a new subject fighting for timetable space. They treat it as a tool students use to *demonstrate* learning they're already doing.

A sustainability-focused school themes the build day around "make an app that helps the planet." A school running a year-long inquiry maps the theme onto it. A STEAM-heavy school points students at "build a tool that solves a real problem you've noticed at school." The AI is the instrument; your curriculum is the music.

There's a future-readiness story in here too, and it's one you can tell parents and prospective families honestly. A school that teaches students to build with AI — rather than banning it or ignoring it — is visibly preparing them for the world they'll graduate into. That's a real differentiator at Open Day, and it's true, which is the only kind of claim worth making.

## A note on cost

Schools ask about pricing early, and fairly. Programmes are scoped per school — a single year-group workshop and a whole-school multi-day residency are very different things — so the format and price are laid out in a written proposal after we talk, not pulled off a shelf.

If you want market context while you're scoping budget, our parent-facing price guides for [Singapore](/blog/ai-classes-cost-singapore-price-guide) and [Hong Kong](/blog/ai-classes-cost-hong-kong-price-guide) give a sense of what well-run, hands-on AI teaching costs to deliver. School programmes are priced differently, but the same principle holds: you're paying for real teaching, real tools, and a real shipped result — not a slide deck.

## Where to start

The lowest-commitment way to find out if this fits your school is a [discovery call](/schools). We'll ask about your priorities and your worries, you'll ask us anything you like, and we'll follow up with a written proposal a few days later. No pressure, and no slides on that call.

Or browse the full menu of [school programmes](/schools) — whole-school hackathons, weekly afterschool, holiday camps, and teacher PD — to see what shape might fit your campus. If you're weighing this for the adult or corporate side of your community too, [/companies](/companies) covers that.

You don't have to have AI figured out before you start. None of us did. You just have to decide your students are better off learning it *with* you in the room than without you.

— *Mr. Brown*


---

## What is Claude Code? A 4-minute primer for non-coders

> Claude Code is the AI coding tool that has changed what a non-engineer can build in a weekend. Here's what it is, how it differs from ChatGPT and Cursor, and how to start using it today.

Published: 2026-05-31 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/what-is-claude-code

A student asked me last week, with genuine confusion:

> "Mr. Brown, everyone keeps saying 'just use Claude Code.' But I already use Claude in the chat window. What's the difference? Is Claude Code a different thing?"

Yes — and the difference is the whole reason a 12-year-old in my Saturday class can ship a working web app before lunch.

Here's the honest version, in plain English.

## The 10-second definition

**Claude Code is the AI coding tool from Anthropic that doesn't just *suggest* code — it writes it, runs it, fixes it, and ships it for you.** You tell it what you want in normal English. It does the rest. It's not a chat window pretending to code. It's an agent that lives on your computer and builds things.

That's the whole thing.

## How it's different from the AI tools you already know

There are roughly four flavours of AI tools that touch code right now, and most people confuse them. Quick map:

★ ChatGPT / Claude (the chat apps)

A chat window. You ask it to write code, it gives you code in a box, you copy-paste it somewhere else, you try to run it. If something breaks, you go back and ask again. There's a lot of human shuffling between you and a working app.

★ GitHub Copilot

Autocomplete for engineers who already write code. Sits in your editor, suggests the next line as you type. Useful if you already know what you're doing. Doesn't help if you don't.

★ Cursor

A whole code editor (it's a fork of VS Code) with AI baked in. Much more powerful than Copilot, but still designed around the idea that a human is the primary driver of the code. You're in the editor, AI is your assistant.

★ Claude Code

An *agent* — meaning it can execute. You give it a goal in normal English. It picks the files, writes the code, runs the code, sees what broke, fixes it, and keeps going until the goal is met. You don't open an editor. You don't paste anything anywhere. You describe, then watch.

For a non-coder, this is the difference between *helping a chef cook* and *ordering room service*. Both are useful. Only one lets someone with zero kitchen skills end up with dinner.

## What you can actually build with it

Three real examples from classes I've run this year, in roughly increasing complexity:

1. **A personal homework helper** that quizzes you on terms you upload from a photo of your textbook. Built by a 13-year-old. ~2 hours.
2. **An internal team dashboard** that pulls Stripe data and shows the metrics your CEO actually asks about. Built by a marketing lead, zero coding background. ~4 hours.
3. **A small App Store iPhone app.** Yes, really — I [built one myself](/blog/30-day-iphone-app-with-ai) in a month with Claude Code as the primary tool. Whole thing is in [vibe coding](/blog/what-is-vibe-coding) territory.

The pattern: you're not learning to write code, you're learning to *direct* code. That skill scales from "weekend project" to "real shippable software" remarkably fast.

## Who Claude Code is actually for

If you recognise yourself in any of these, Claude Code is the tool you've been waiting for:

- **The founder who keeps saying "if only I could build a prototype myself."** You can now. In an afternoon.
- **The marketer or ops person** who's been describing the same internal tool to engineering for two years and never getting it. Build it yourself this Saturday.
- **The teacher** who keeps thinking *if only this lesson had an interactive piece*.
- **The student** (10+, in our experience) who wants to build a real, deployed website with their name on it.

The skill it rewards is *clear thinking* and *taste*. The skill it removes is *remembering syntax*. If you're good at the first two, you're going to be dangerous with this tool.

## How to start this weekend (3 steps)

If you want to try Claude Code with zero prior coding experience, here's the simplest path:

1. **Get an Anthropic account at [claude.com](https://claude.com).** The free tier is fine to start. Claude Code itself runs as a terminal app on your computer — install instructions are at [anthropic.com/claude-code](https://www.anthropic.com/claude-code).
2. **Pick one specific small thing you want to build.** Not "an app for my whole business." More like: "a one-page web tool that does *one specific thing* I keep doing manually." A unit converter. A daily-standup-question randomiser. A pet-name generator. Make it tiny.
3. **Describe it like you're briefing a really fast intern.** "I want a single-page website that takes a URL as input and returns the word count. Make it look clean. Deploy it so I can share the URL." That sentence is enough for Claude Code to actually build it.

Then iterate. Don't rewrite from scratch — say what's bad. "The button colour is ugly." "I want the result to be huge." "Add a copy-to-clipboard button." Each round, Claude Code edits the right files and reloads. You ship.

Total time for your first end-to-end build: 1–2 hours, give or take. Less if you start small.

## What we teach about Claude Code at Pathwise

Claude Code is the primary tool we teach in our [Idea to App program](/programs/vibe-code) — for both [young builders](/young-builders) (ages 10–17) and [adults](/adults). Every cohort, every learner walks out with a deployed app they built with Claude Code as their main builder.

(Full disclosure: Claude Code is also one of our partners — listed on the homepage. We use it because it's the best tool for what we teach, and the partnership grew from that, not the other way around.)

If you want to learn this in a small room with a teacher who builds with Claude Code daily — [browse the next cohort dates](/classes) or [drop us a line](/contact).

The hardest part of learning Claude Code isn't Claude Code. It's deciding what you want to build.

— *Mr. Brown*


---

## How much do AI classes cost in Hong Kong? A 2026 parent's price guide (HKD)

> Realistic HKD pricing ranges for AI classes for kids in Hong Kong — online cohorts, in-person workshops at Central/Causeway Bay/Kowloon venues, holiday camps, and what you're actually paying for at each tier.

Published: 2026-05-27 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-cost-hong-kong-price-guide

A Hong Kong parent messaged me last week — the same question I get more than any other from this side of the region:

> "What should I expect to pay for an AI class for my child in Hong Kong? Some of the ones I'm seeing are HKD 200 a session. Others are HKD 1,800. They both call themselves 'AI classes.' What gives?"

Here's the honest answer, in HKD, for parents in Hong Kong in 2026. Real ranges, what you're paying for at each tier, and the red flags to watch for.

## The short version

For a real, hands-on AI class for kids in Hong Kong in 2026, expect:

- **Online cohorts**: HKD 350–600 per session
- **In-person workshops** (Central / Causeway Bay / Kowloon venues): HKD 900–1,400 per session
- **Holiday camps** (full day, Easter / summer / Christmas): HKD 3,500–7,000 for the week
- **One-on-one tutoring**: HKD 1,200–3,000 per hour (rare; usually overkill)
- **Free or under HKD 200 per session**: nearly always volunteer instructors or a marketing funnel for something else

That's the band of legitimate, well-run AI programs in Hong Kong right now. If a class is priced significantly above or below those ranges, ask why.

## What you're actually paying for

The price spread isn't arbitrary. It tracks three things:

### 1. Instructor quality

The biggest line item. A good AI instructor in Hong Kong — someone who has actually shipped software, kept up with tools that change every six weeks, and can teach kids in English without losing them — earns HKD 500–900/hour or more in the broader Hong Kong market. Two- to four-hour classes with that calibre of instructor land at HKD 900–1,400 per session naturally.

Volunteer instructors and undergraduate teaching assistants cost much less. Some are excellent. Many treat the class like a homework session — open Scratch, follow the worksheet, sit at the back. You won't know which kind you got until you're three sessions in.

### 2. Class size

A class of 8 students is fundamentally different from a class of 25. With 8, the instructor sees every screen, debugs every bug, gives real feedback on every build. With 25, you get lecture-then-confusion: a 20-minute talk followed by 25 kids stuck on different problems with one adult.

A HKD 500 session in a class of 25 = roughly HKD 20 of instructor attention per kid. A HKD 1,000 session in a class of 8 = HKD 125. The smaller class is a dramatically better deal for your child even at the higher headline price.

### 3. Tools, deployment, and operations

Real AI classes provide accounts to Claude, ChatGPT, Midjourney, Figma, deploy targets, and snacks. That stack costs the operator real money per cohort. Bargain-basement programs hand kids a free ChatGPT account and a Scratch link. You can spot the difference inside one session.

## The pricing bands explained

### Under HKD 200 per session (or free)

These are almost always one of three things:

1. **A subsidised intro/trial** designed to upsell into something paid later
2. **A volunteer-led community program** (a university outreach session, a free library event)
3. **A marketing event** for a coding bootcamp, school chain, or product

None of these are bad on their own — a free trial is a great way to test fit before committing. But you should know what you're getting.

### HKD 200–500 per session

Usually online, group-based, classroom-style. Often run by a coding school adding "AI" to existing curriculum. Quality varies wildly. The good ones have a teacher you can ask questions to in real-time; the cheaper ones are prerecorded with light async feedback. Worth doing a free trial before signing up for a full term.

### HKD 500–1,000 per session (the realistic floor for a good online class)

This is where serious, hands-on, small-cohort *online* classes in Hong Kong sit. You're paying for: a real teacher, real cohort (≤10 students), real tools, real builds, real deployment. If you only have budget for one AI class for your child this term and you're in Hong Kong, this is the band to aim for online.

### HKD 900–1,400 per session (the realistic floor for in-person)

This is where serious, hands-on, small-cohort *in-person* classes in Hong Kong sit. Reputable in-person programs — Pathwise included — land in this range. The premium over online is real: physical room energy, full instructor attention, and a venue cost on top of everything else.

### HKD 1,400–2,500 per session (premium small-group)

Smaller cohorts (3–6 students), more 1:1 attention, possibly with a "name" instructor. Good for kids who need more pace control — either much faster or much slower than a standard cohort. Premium pricing reflects the smaller class size, not necessarily better content.

### HKD 2,500+ per session (one-on-one tutoring)

Almost always 1:1. Useful if your child has very specific goals (e.g. building one ambitious project, or prepping for an academic competition). Overkill for most kids — the energy of a small group is part of the learning, not a distraction from it.

## Holiday camp pricing

Multi-day camps during Hong Kong school holidays (Easter, summer, Christmas) are priced as packages, not per session:

- **3-day half-day camp**: HKD 2,000–3,500
- **5-day full-day camp**: HKD 3,500–7,000
- **Weekend intensive** (2 days): HKD 2,000–4,000

These can be a great deal if your child can sustain a full day. They get more momentum and ship something more substantial than a single Saturday workshop. Watch out for camps that are mostly "screen time + a worksheet" — the good ones ship something real by the end of the week. See the full [AI holiday camps guide for Singapore & Hong Kong](/blog/ai-holiday-camps-for-kids-singapore-hong-kong) for camp formats.

## Red flags in pricing

A few things to watch for when comparing Hong Kong AI classes:

- **"Up to 70% off!" discounting** — usually inflating the base price to make the discount look bigger. Compare the actual paid price to others, not the percentage.
- **No clear class size disclosed** — operators avoid this when classes are big. Ask directly: "How many students per instructor?"
- **No deployed artifact at the end** — if there's no live URL / shipped project, you're paying for theory, not skill.
- **Charging premium prices for "AI awareness" content** — discussions about ethics and the future of AI are valuable, but they're an article, not a HKD 1,200/session class.
- **Long-term packages with no trial** — any legitimate program will let you book a single session or have a free trial. If they only sell 12-week prepaid packages, that's a sales tactic, not a teaching tactic.
- **Cantonese-only delivery when your kid is at an international school** — confirm the language of instruction before booking. Most reputable HK providers teach in English with bilingual learners welcome.

## Where Pathwise sits

For context: each Pathwise session in Hong Kong caps at around 8 students, runs on real professional tools (Claude, Figma, Midjourney), and every learner walks out with a real, deployed web app. We sit in the realistic band for what hands-on, small-cohort AI classes taught by a working teacher cost to run sustainably in Hong Kong.

We aren't the cheapest — and we don't try to be. If price is the only factor, there are cheaper options. If you want your child to walk out of a HK venue with something they actually built and a URL to share with grandparents in another country, this is roughly where the market is.

For more on what to look for inside a class — beyond price — see [AI classes for kids in Hong Kong: a parent's guide](/blog/ai-classes-for-kids-hong-kong). For the Singapore-side equivalent of this price guide, see the [SGD version here](/blog/ai-classes-cost-singapore-price-guide).

## So what should you do?

Three practical steps:

1. **Decide your format first** (online vs in-person, weekly vs camp). Each has its own price band; don't compare across.
2. **Compare like-for-like**. Three HKD 600 sessions of 8-kid in-person is not the same as three HKD 300 sessions of 25-kid online — even if the per-session price is similar.
3. **Always ask for a single-session trial** before a full term. Real programs offer this. Watch your child's eyes 30 minutes in. If they're building, the price is irrelevant. If they're staring at a slide deck, walk away regardless of price.

If you want to see what an in-person Pathwise session in Hong Kong looks like before booking, the [classes & schedule](/young-builders#schedule) page has the next two months of dates. Or [WhatsApp us](https://wa.me/6580261562) and we'll send you a recap from the most recent cohort.

— *Mr. Brown*


---

## AI classes for adults in Singapore: a non-coder's guide for 2026

> What to actually look for in an AI class for adults in Singapore — formats, who they're for, the red flags to skip, and what to expect to pay. From a teacher who runs them.

Published: 2026-05-24 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-for-adults-singapore

The best AI classes for adults in Singapore in 2026 are **build-based, small-cohort, and taught by someone who actually ships with AI** — not slide-deck explainers, not "AI awareness" panels, not 50-person webinars. Look for a class where you leave the room with something you built that week (a deployed web app, a working agent, a designed brand). Expect SGD 80–200 per session in-person, SGD 60–100 online; one-day intensives SGD 300–600. If the deliverable is "skills" instead of "your own project on the internet," skip it.


A friend in her mid-30s, two career changes deep, asked me over coffee:

> "Everyone keeps telling me to 'learn AI.' I don't want a certificate. I don't want to memorise prompt frameworks. I want to actually build something. What kind of class do I sign up for in Singapore?"

She'd already been through a four-hour webinar that turned out to be a sales pitch, a "ChatGPT for professionals" course that was mostly a list of prompts, and a free meetup where someone read slides about the "future of work." Three swings, three misses.

Here's the honest version — what good AI classes for adults in Singapore actually look like in 2026, who they're for, and what to expect to pay.

## What "AI classes for adults" actually means in 2026

The market has split into three flavours, and only one of them teaches a skill that compounds:

1. **Prompt classes** — you learn templates and frameworks for talking to ChatGPT and Claude. Useful background, but it's typing class for 2026: necessary, not sufficient. You'll forget most of the frameworks within a month.
2. **Build classes** — you use AI to *make real things*. A working web app. A live AI agent that does a job for you. A designed brand system. The skill is judgement — knowing what to ask for, knowing when the AI is wrong, knowing what to keep.
3. **AI strategy / leadership classes** — discussions about org change, governance, the future of work. Useful if you're a CEO planning policy. Mostly noise if you actually want to make something yourself.

If you're an adult who wants to *do* something with AI rather than *talk about* AI, the build classes are the only category worth your time. Everything else is reading material in expensive clothing.

## 5 markers of a good AI class for adults

After running adult cohorts in Singapore for the last couple of years, here's what I'd check before paying for anything.

### 1. You ship a real thing — that session

Not "by the end of an eight-week course." That session. If a class promises "exposure to concepts" and the deliverable is a certificate or a slide deck, walk away. The shipped artifact — a URL, a working file, an agent that does its job — is what proves the learning landed.

★ Test the brochure

Search the syllabus for the words "deployed," "live," "URL," or "your project." If they're not there, ask directly: *"What do I walk out of session one with?"* If the answer involves the word "framework" or "certificate," that's your answer.

### 2. Real tools, real accounts

Claude. Claude Code. ChatGPT. Cursor. Replit. Midjourney. Figma + AI. The same tools working professionals use every day. Some providers use sandboxed "intro" tools so they don't have to deal with billing — the skill you learn doesn't transfer. Look for classes that use the actual paid-tier professional tools on actual accounts.

### 3. Small cohorts — under 12 adults per instructor

You cannot ship a working app in a room of 30 with one instructor. The format requires real attention per person: someone looking at *your* idea, debugging *your* code, telling you why *your* design isn't quite landing. Anything above ~12 becomes lecture-then-confusion. The good classes in Singapore cap below this number on purpose.

### 4. The teacher should also be a builder

Ask: *"Has the instructor actually shipped software, run a business, or worked in the role I want to be in — using AI?"* Both academic and practitioner instructors exist. Only the practitioner will give you instincts.

Look at their portfolio. Public projects, launched apps, an app store listing, a real consulting record. If it's all talks and articles, you're getting theory.

### 5. Peer cohort that's a fit for your goal

Adult AI classes pull a wider mix than most education: a 28-year-old founder, a 45-year-old marketer, a 60-year-old senior partner figuring out what to do next. That mix is part of the value — but only if the class is genuinely small-group, with real time for discussion. If the cohort is anonymous to each other by week three, you've been sold a webinar, not a cohort.

## Who AI classes for adults are actually for

If you recognise yourself in any of these, you're in the target audience:

- **Career pivoters.** You're 30–50, considering a move into AI-adjacent work (product, design, ops, consulting). You need to build a real thing you can show, not just say "I learned about AI."
- **Founders and would-be founders.** You keep saying *"if only I could build a prototype myself"*. With vibe coding, you can — in a weekend.
- **Marketers, ops people, and operators.** You want to automate the workflow you've been doing manually for years. An AI agent in your pocket is now a 4-hour weekend project, not a six-month engineering ticket.
- **Designers.** You want to direct AI design tools with taste, not be afraid of them. The good classes treat Figma + AI and Midjourney as instruments, not magic.
- **Curious adults.** You don't have a use case yet. You just want to feel competent. That's a fine reason — the build itself often reveals the use case.

## Red flags to skip

In rough order of severity:

- **"AI awareness" with no building.** Theory-only formats treat you as an audience, not a maker. Skip.
- **Cohorts above 30.** The format doesn't survive scale. You'll get a webinar with breakout rooms.
- **Prerecorded modules sold as a "course."** If there's no live instructor you can ask "wait, why did it do that?", you're buying a Udemy course at a premium.
- **No deliverable.** If you can't articulate what you'll have at the end — a URL, a file, a working agent — there is no end.
- **"Certified" — by whom?** Adult AI certifications in 2026 are nearly meaningless. The portfolio is the certification.
- **Pricing under SGD 30 per session.** Either subsidised intro, or volunteer instructors. You usually get one good one and a lot of bad ones.

## What it should cost in Singapore

Rough ranges I've seen for legitimate, hands-on adult AI classes in Singapore:

- **Online weekly cohorts**: SGD 60–100 per session
- **In-person half-day workshops** (Singapore CBD): SGD 150–250 per session
- **Weekend intensives (1–2 days)**: SGD 300–600 total
- **Corporate team workshops** (custom-scoped, on-site): SGD 2,000–6,000 per day depending on team size and depth
- **Free or under SGD 30**: usually a marketing event for a bootcamp or consultancy — fine for sampling, not for skill-building

You're paying for instructor quality, cohort size, and the fact that good live AI instructors in Singapore are scarce. Above SGD 300 per session, you should be getting either a name instructor, an exceptionally small group (≤6), or unusually high-touch follow-up.

## Common questions

**I haven't coded in years. Or ever. Is this for me?**
Yes. Most adult cohorts at Pathwise are non-engineers — operators, marketers, founders, the occasional doctor. The whole point of vibe coding is that you don't write syntax. You describe what you want, the AI builds it, you direct and refine. If you can write a clear email, you can do this.

**How much time per week do I need?**
Depends on the format. A one-Saturday workshop is exactly that — show up, build, leave with a URL. A four-week cohort is usually 2–3 hours of live class plus optional homework. We're upfront about which format suits which goal.

**What tools will I actually learn?**
Claude (and [Claude Code](/blog/what-is-claude-code), which is the one I personally use most). ChatGPT. AI design tools (Figma + AI, Midjourney). Deployment platforms like Replit and Vercel. The same stack working pros actually use day-to-day — see the [vibe coding primer](/blog/what-is-vibe-coding) for a longer walkthrough.

**Is there an adults-only cohort, or am I in a room with teenagers?**
Adults-only. Same teacher, same builds, different pace and content. Adult cohorts skew more conceptual conversation; the kids' cohorts skew more "did you SEE what I made?" energy.

**What happens after the class ends?**
You leave with a deployed project and the muscle memory to make a second one in half the time. Many of our adult alumni come back for a second cohort six months later with a more ambitious idea. The skill compounds.

## What I'd actually do

If I were the friend who emailed me, I'd do this in order:

1. **Try one thing this weekend** before paying for anything. Open [Claude](https://claude.com), describe one small tool you keep wanting (a unit converter, a meeting-notes summariser, a tiny personal dashboard) and see how far you get in an hour. If you feel the giddy "wait, I made this?" — you're in the target audience.
2. **Book one in-person session.** A half-day with a real instructor in the room beats four weeks of online for the first taste. You either love it or you don't, and you'll know by lunch.
3. **Then commit to a longer cohort** if step 2 lands. That's where the skill actually compounds.

We run [adult cohorts in Singapore](/adults) — small, in-person at JustCo Marina Square or online for anyone in APAC. Each session ends with a real, deployed project you keep. See the [upcoming schedule](/adults#schedule) or browse the [programs we teach](/programs/vibe-code). If you want to ask anything specific before booking — [drop us a line](/contact).

The hardest part of learning AI as an adult isn't the AI. It's giving yourself permission to be a beginner at something again.

— *Mr. Brown*


---

## AI classes for kids in Hong Kong: a parent's guide for 2026

> What AI classes for kids in Hong Kong actually teach, what to avoid, age-appropriate formats, English-medium options, and how much to expect to pay (HKD).

Published: 2026-05-15 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-for-kids-hong-kong

The best AI classes for kids in Hong Kong (ages 10–17) are **build-based, not theory-based**: your child should walk out with a real, deployed web app at a live URL. Look for small cohorts (~8 students), real professional tools (Claude, Figma, Midjourney), English-medium delivery, and an instructor who actually ships software. Online weekly cohorts typically run HKD 350–600/session, in-person workshops HKD 900–1,400/session, and school-holiday camps HKD 3,500–7,000 for the week. If a class ends in a certificate instead of a URL, skip it.


Hong Kong parents ask me a sharper version of the question I get in Singapore:

> "There are dozens of robotics and coding centres rebranding as 'AI for kids'. Which ones actually teach AI, and which are STEM camps with a new sticker?"

Fair. Hong Kong has a deep STEM-tutoring market, and a lot of it has simply relabelled. Here's the parent's guide for a 10- to 17-year-old in Hong Kong in 2026 — what to look for, what to skip, and what it should cost in HKD.

## What "AI classes for kids" should mean in 2026

There are three flavours on the market, and only one builds a durable skill:

1. **Prompt classes** — kids learn to talk to ChatGPT/Claude. Useful, but it's typing class for 2026: necessary, not sufficient.
2. **Build classes** — kids use AI to *make real things*: web apps, designs, working agents. This is where the actual skill lives.
3. **AI literacy / ethics classes** — discussions about bias and deepfakes. Important, but a child who has never built with AI isn't equipped to think critically about it. Theory without practice is just opinion.

If you're choosing, build-based wins by a wide margin. A child who has shipped a working web app has had a fundamentally different experience to one who has only used ChatGPT for homework.

## What good AI classes for kids in Hong Kong look like

### 1. Every student leaves with a live URL

The single most important filter. If the syllabus doesn't contain "deployed", "live", or "your child's own app", ask directly: *"What does my child walk out with on day one?"* If the answer is a certificate, walk away.

### 2. Real tools, not "kid versions"

Claude, ChatGPT, Midjourney, Figma — the same tools working professionals use, with adult supervision instead of artificial guardrails. Kids handle real tools better than most adults.

### 3. English-medium, with bilingual learners welcome

Most international-school students and adult learners in Hong Kong are fluent in English. Good providers teach in English without requiring Cantonese or Mandarin, while welcoming bilingual learners. Confirm the language of instruction before booking.

### 4. Small cohorts (~8 students)

You cannot ship a working app in a room of 30 with one instructor. The format requires real attention per child — debugging their problem, reviewing their design. Anything over ~10 per instructor becomes lecture-then-confusion.

### 5. The teacher should also be a builder

Ask: *"Has the instructor actually shipped software with AI, or only read about it?"* Look at their portfolio — launched apps, public projects, a working agent. If yes, they teach in a way that compounds.

## Red flags to skip

- **"AI awareness" with no building** — treats your child as an audience, not a maker.
- **Large class sizes (20+)** — the format doesn't survive scale.
- **Worksheets and multiple-choice "AI safety" modules** with no actual AI work — usually old curriculum with a new label.
- **No mention of what the child takes home** — if the deliverable is "skills" not "their own working project", it's vapor.

## What ages benefit most

- **Ages 10–11**: Best in short, in-person, parent-present formats. Huge imagination, needs a teacher in the room.
- **Ages 12–14**: The strongest cohort — old enough to self-direct, young enough not to be cynical. Most dramatic builds happen here.
- **Ages 15–17**: Handle pro-level workshops at adult pace. Treat them like junior developers.

For more on the age question, see [what age a child should start AI classes](/blog/what-age-should-a-child-start-ai-classes).

## What AI classes for kids cost in Hong Kong

Rough ranges for legitimate programs, in HKD:

- **Online weekly cohorts**: HKD 350–600 per session
- **In-person workshops** (Central / Causeway Bay / Kowloon venues): HKD 900–1,400 per session
- **Holiday camps** (full day, Easter / summer / Christmas): HKD 3,500–7,000 for the week
- **Whole-school programs / hackathons**: scoped per school

You're paying for instructor quality, class size, and tools. Below HKD 200/session you're getting volume or volunteer instructors.

## Pathwise in Hong Kong — at a glance



## Where to go from here

Two paths most Hong Kong parents take:

1. **Try it at home first.** Read the [step-by-step guide to introducing AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old) — three things to try this weekend before paying for anything.
2. **Join a structured class.** We run [after-school and weekend classes for ages 10–17](/young-builders) in Hong Kong and online — see the [Hong Kong page](/hong-kong) for venues and dates. Every learner leaves with a real, live web app and a URL.

Whichever path you pick: look for the live URL at the end. That's the test that matters. Everything else is decoration.

— *Mr. Brown*


---

## AI classes vs coding classes for kids: which should your child do in 2026?

> A teacher's honest comparison of AI classes and traditional coding classes for kids — what each teaches, which age each suits, and how to choose for a 10–17-year-old.

Published: 2026-05-15 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-vs-coding-classes-for-kids

For most kids aged 10–17 in 2026, **start with an AI build class, not a traditional syntax-first coding class**. AI classes get a child to a real, deployed app in days, which builds motivation and judgement; traditional coding teaches durable fundamentals but is slower and loses many kids before they ship anything. The best answer for many families is **AI-first, then add coding fundamentals later if the child is hooked**. If your child already loves coding, keep going — add AI tools on top.


This is now the most common question I get from parents:

> "Should my child do an AI class, or a proper coding class? Isn't coding the 'real' skill?"

It's a genuinely good question, and the honest answer has changed in the last two years. Here's how I'd think about it for a 10- to 17-year-old.

## What each one actually teaches

**Traditional coding classes** teach syntax-first fundamentals: variables, loops, functions, data structures, often in Python, Scratch, or JavaScript. The promise is durable understanding of how software works under the hood.

**AI build classes** teach a child to describe what they want, direct an AI tool (Claude, Cursor, Figma) to produce it, read and debug the result, and ship it live. The promise is agency: a real thing on the internet, fast.

Both are legitimate. They build different muscles.

## The honest comparison


**Traditional coding class**

- **Time to first real, shipped thing:** weeks to months
- **What builds:** deep fundamentals, patience, debugging from first principles
- **Risk:** many kids quit before they ship anything they're proud of
- **Best for:** kids who already enjoy puzzles and tolerate slow progress



**AI build class**

- **Time to first real, shipped thing:** the first session
- **What builds:** judgement, taste, prompting, the confidence that comes from shipping
- **Risk:** without good teaching, a child can stay shallow and never learn *why* things work
- **Best for:** most kids 10–17, especially those who've never finished a coding course


The decisive factor for most families is **motivation**. A child who deploys a working app in their first session almost always wants to go deeper — and *that* is when fundamentals become worth learning, because now they have a reason to care.

## Isn't coding becoming obsolete?

No — but the *entry point* has changed. Professional engineers still need fundamentals. But in 2026, the fastest way to *get a child to want* those fundamentals is to let them build something real with AI first, then work backwards into how it actually works. Syntax-first, for most kids, front-loads the boring part and loses them before the payoff.

Think of it like music: you don't start a 10-year-old on six months of theory. You get them playing a song they love in week one, then teach theory once they're hooked.

## How to choose, by situation

- **Your child has never finished a coding course / gets bored fast:** Start with an AI build class. Ship something real, build the appetite.
- **Your child already loves coding:** Keep going — and add AI tools on top. They'll move faster than ever.
- **You want the "real" computer-science path:** Do AI-first for motivation, then add a structured fundamentals course within 6–12 months.
- **Ages 10–12 specifically:** AI build classes win almost every time — the shipped result sustains attention that syntax drills can't.

For more on the age question, see [what age a child should start AI classes](/blog/what-age-should-a-child-start-ai-classes).

## What a good AI build class looks like

Same filters as any AI class for kids: every student leaves with a **live URL**, real professional tools, small cohorts (~8 students), and an instructor who actually ships software. If you want the full checklist, read [AI classes for kids in Singapore: what to look for](/blog/ai-classes-for-kids-singapore) (the same principles apply in Hong Kong and online).

## Pathwise at a glance



## Where to go from here

If you've decided AI-first makes sense, we run [build classes for ages 10–17](/young-builders) in Singapore, Hong Kong, and online — every learner ships a real, live web app in their first session. Or [book a class](/classes) and see the upcoming schedule.

The goal isn't "AI instead of coding". It's getting your child to *want* to go deep — and in 2026, building something real with AI is the fastest door into that.

— *Mr. Brown*


---

## AI holiday camps for kids in Singapore & Hong Kong: 2026 guide

> A parent's guide to AI school-holiday camps for kids (10–17) in Singapore and Hong Kong — formats, what they build, when camps run, and what to expect to pay.

Published: 2026-05-15 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-holiday-camps-for-kids-singapore-hong-kong

A good AI holiday camp gets a child (ages 10–17) from idea to a **real, deployed web app in one week** — not slideshows about robots. In Singapore, camps run during the June, September, and December school holidays; in Hong Kong, during Easter, summer, and Christmas breaks. Expect roughly **SGD 600–1,200 (~HKD 3,500–7,000) for a full-day week**. The test of a real camp: every child leaves with a live URL they can show, not a printed certificate.

Pathwise runs full-day holiday camps in Singapore and Hong Kong — see the [Singapore](/singapore) and [Hong Kong](/hong-kong) pages for dates.


School holidays are when a lot of parents finally try an AI class — there's time, and a week-long camp is a low-commitment way in. But "AI camp" now covers everything from serious build programs to repackaged robotics day-care. Here's how to tell them apart.

## What a good AI holiday camp actually does

In one week, a real camp takes a child from *idea* to *shipped*. By Friday, a 12-year-old should be able to send you a link and say "I made this." The week usually looks like:

- **Day 1:** Pick an idea, learn to direct the AI, get a first version live.
- **Days 2–3:** Build the real features. Hit bugs, learn to debug with AI.
- **Day 4:** Design and polish — make it something they're proud to share.
- **Day 5:** Ship the final version, demo it to the group, take the URL home.

If the brochure describes "exposure to AI concepts" and "fun activities" but never mentions a thing the child takes home and keeps, it's day-care with a keyboard.

## When camps run

**Singapore** — school-holiday camps run during:

- **June** (mid-year break) — the biggest season
- **September** (Term 3 break) — shorter, fills fast
- **December** (year-end) — popular, books early

**Hong Kong** — camps run during:

- **Easter break** (spring)
- **Summer** (July–August) — the biggest season
- **Christmas** (December)

Dates are typically published 6–8 weeks ahead. Popular weeks fill, so the practical advice is: decide the season now, book when dates open.

## What it costs

Full-day camps for a week, for legitimate programs:

- **Singapore:** SGD 600–1,200 for the week
- **Hong Kong:** HKD 3,500–7,000 for the week

You're paying for full days, small groups, real tools, and an instructor who can get every child to ship. Below those ranges you're usually getting large groups or volunteer helpers; the tell is the group size and whether every child leaves with a live project.

## Picking the right week for your child's age

- **Ages 10–12:** Full-day camps work well *if* the group is small and there's a teacher in the room. The shipped result sustains attention a worksheet can't.
- **Ages 12–14:** The ideal camp age — enough independence to build fast, enough structure to stay on track.
- **Ages 15–17:** Look for an "intensive" or advanced track; they'll move quickly and can take on agents or more ambitious builds.

For the full age-by-age breakdown, see [what age a child should start AI classes](/blog/what-age-should-a-child-start-ai-classes).

## Holiday camp vs weekly classes — which first?

A camp is the best *first* taste: one intense week, a finished project, a clear signal of whether your child is hooked. If they come home wanting more, weekly cohorts go deeper over time. Many families do a holiday camp first, then continue with [after-school or weekend classes](/young-builders).

## Pathwise camps at a glance



## Where to go from here

We run full-day AI holiday camps for ages 10–17 in Singapore and Hong Kong — every camper ships a real, live web app by the end of the week. Check the [Singapore page](/singapore) or [Hong Kong page](/hong-kong) for upcoming camp dates, or [book a class](/classes) to see the live schedule.

The one test that matters: by Friday, can your child send you a link to something they built? If yes, it was a real camp.

— *Mr. Brown*


---

## What age should a child start AI classes?

> A teacher's honest answer on the right age to start AI classes — what 10–12, 12–14, and 15–17 year olds can each do, and what to do before age 10.

Published: 2026-05-15 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/what-age-should-a-child-start-ai-classes

**Most children are ready for a real AI build class at age 10.** Ages 10–12 do best in short, in-person, parent-present formats; ages 12–14 are the strongest cohort and can self-direct; ages 15–17 can handle pro-level workshops at adult pace. Under 10, skip formal classes — do guided exploration at home instead. The readiness test isn't age, it's whether the child can read, type a sentence, and stay with a task for ~20 minutes.


"What age should my child start AI classes?" is one of the two questions I'm asked most (the other is what it costs). Here's the honest answer from teaching hundreds of young builders.

## The short answer: around age 10

For a structured, build-based AI class, **10 is the practical floor** for most children. By 10, a typical child can read instructions, type a sentence, and stay with a task long enough to see a result. Those three things — not coding knowledge — are what actually gate readiness.

The real readiness test:

- Can they **read** a short instruction on screen?
- Can they **type** a sentence describing what they want?
- Can they **stay with** one task for ~20 minutes?

If yes, they're ready to build with AI — regardless of whether they've ever coded.

## What each age band can do

### Ages 10–12 — the curious beginners

Huge imagination, low cynicism, high willingness to try. They thrive in **short, in-person, parent-present** formats with a teacher who can lean over and look at their screen. They can absolutely ship a real web app — they just need a person in the room, not a screen, for the first sessions.

### Ages 12–14 — the strongest cohort

Old enough to self-direct, young enough not to be cynical about it. This is where I've seen the most dramatic builds. They can handle weekly online cohorts or in-person workshops and will often outpace expectations within a few sessions.

### Ages 15–17 — junior builders

Can handle pro-level workshops at adult pace. Treat them like junior developers, not students. The good ones quietly outpace their parents within a few sessions and can move into agents and more advanced builds.

For a deeper age-by-age breakdown of formats and what to look for, see [AI classes for kids in Singapore: what to look for](/blog/ai-classes-for-kids-singapore).

## What about under 10?

Skip formal classes. A 7- or 8-year-old doesn't need a structured AI cohort — they need **guided exploration at home** with a parent: making images, asking questions, drawing with AI, all with an adult steering. Our [guide to introducing AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old) works as a gentler at-home starting point for younger kids too.

The mistake is rushing a 7-year-old into a class built for 12-year-olds. Readiness compounds — a motivated 11-year-old beats a pushed 8-year-old every time.

## Is it ever too late to start?

No. A 16-year-old starting from zero is in a *great* position — more focus, faster reading, more patience for iteration. There is no "missed the window" with AI. The tools reward judgement, and judgement keeps developing through the teens.

## Pathwise at a glance



## Where to go from here

If your child is 10 or older and passes the read/type/stay test, they're ready. We run [build classes for ages 10–17](/young-builders) in Singapore, Hong Kong, and online — every learner ships a real, live web app in their first session. [Book a class](/classes) to see upcoming dates, or start at home with the [parent's intro guide](/blog/how-to-introduce-ai-to-your-10-17-year-old).

— *Mr. Brown*


---

## How much do AI classes cost in Singapore? A 2026 parent's price guide

> Realistic SGD pricing ranges for AI classes for kids in Singapore — online cohorts, in-person workshops, holiday camps, and what you're actually paying for at each price point.

Published: 2026-05-14 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-cost-singapore-price-guide

A parent emailed me last week asking the question I get more than any other:

> "What should I expect to pay for an AI class for my child in Singapore? Some of the ones I'm seeing are SGD 30 a session. Others are SGD 300. They both call themselves 'AI classes.' What gives?"

Here's the honest answer I gave her, written out properly. Real SGD ranges for AI classes in Singapore in 2026, what you're paying for at each tier, and the red flags to watch for.

## The short version

For a real, hands-on AI class for kids in Singapore in 2026, expect:

- **Online cohorts**: SGD 60–100 per session
- **In-person Saturday workshops**: SGD 150–200 per session
- **Holiday camps** (full day, 4–5 days): SGD 600–1,200 for the week
- **One-on-one tutoring**: SGD 200–500 per hour (rare; usually overkill)
- **Free or under SGD 30 per session**: nearly always volunteer instructors or a marketing funnel for something else

That's the band of legitimate, well-run AI programs in Singapore right now. If a class is priced significantly above or below those ranges, ask why.

## What you're actually paying for

The price spread isn't arbitrary. It tracks three things:

### 1. Instructor quality

The biggest line item. A good AI instructor in Singapore — someone who has actually shipped software, kept up with tools that change every six weeks, and can teach kids without losing them — earns SGD 80–150/hour or more in the broader market. Two- to four-hour classes with that quality of instructor land at SGD 150-200/session naturally.

Volunteer or undergraduate instructors cost much less. Some are great. Many treat the class like a homework session — open Scratch, follow the worksheet, sit at the back. You won't know which you got until you're already three sessions in.

### 2. Class size

A class of 10 students is fundamentally different from a class of 30. With 10, the instructor sees every screen, debugs every bug, gives real feedback on every build. With 30, you get lecture-then-confusion: a 20-minute talk followed by 30 kids stuck on different problems with one adult.

A SGD 80 session in a class of 30 = roughly SGD 8 of instructor attention per kid. A SGD 150 session in a class of 10 = SGD 15. The smaller class is the better deal for your child even at higher headline price.

### 3. Tools, deployment, and operations

Real AI classes provide accounts to Claude, ChatGPT, Midjourney, Figma, deploy targets, and snacks. That stack costs the operator real money per cohort. Bargain-basement programs hand kids a free ChatGPT account and a Scratch link.

## The pricing bands explained

### Under SGD 30 per session (or free)

These are almost always one of three things:
1. **A subsidized intro/trial** designed to upsell into something paid later
2. **A volunteer-led community program** (e.g. a polytechnic outreach session)
3. **A marketing event** for a coding bootcamp, school chain, or product
4. **AI-themed theory class** with no actual building — slides and discussion only

None of these are bad on their own — a free trial is a great way to test fit before committing. But you should know what you're getting.

### SGD 30–80 per session

Usually online, group-based, classroom-style. Often run by a coding school adding "AI" to existing curriculum. Quality varies wildly. The good ones have a teacher you can ask questions to in real-time; the cheaper ones are pre-recorded with light async feedback. Worth doing a free trial before signing up for a full term.

### SGD 80–160 per session (the realistic floor for a good in-person class)

This is where serious, hands-on, small-cohort in-person classes in Singapore sit. Reputable in-person programs — Pathwise included — land in this range. You're paying for: real teacher, real cohort (≤10 students), real tools, real builds, real deployment.

If you only have budget for one AI class for your child this year, this is the band to aim for. Two in-person sessions taught well will move the needle more than ten cheap online sessions.

### SGD 160–250 per session (premium small-group)

Smaller cohorts (3–6 students), more 1:1 attention, possibly with a "name" instructor. Good for kids who need more pace control — either much faster or much slower than a standard cohort. Premium pricing reflects the smaller class size, not necessarily better content.

### SGD 250+ per session (one-on-one tutoring)

Almost always 1:1. Useful if your child has very specific goals (e.g. building one ambitious project, or prepping for an academic competition). Overkill for most kids — the energy of a small group is part of the learning, not a distraction from it.

## Holiday camp pricing

Multi-day camps during school holidays (June, September, December) are priced as packages, not per session:

- **3-day half-day camp**: SGD 350–600
- **5-day full-day camp**: SGD 600–1,200
- **Weekend intensive** (2 days): SGD 350–700

These can be a great deal if your child can sustain a full day. They get more momentum and ship something more substantial than a single Saturday workshop. Watch out for camps that are mostly "screen time + a worksheet" — the good ones ship something real by the end.

## Red flags in pricing

A few things to watch for when comparing Singapore AI classes:

- **"Up to 70% off!" discounting** — usually inflating the base price to make the discount look bigger. Compare the actual paid price to others, not the percentage.
- **No clear class size disclosed** — operators avoid this when classes are big. Ask directly: "How many students per instructor?"
- **No deployed artifact at the end** — if there's no live URL / shipped project, you're paying for theory, not skill.
- **Charging premium prices for "AI awareness" content** — discussions about ethics and the future of AI are valuable, but they're an article, not a SGD 200/session class.
- **Long-term packages with no trial** — any legitimate program will let you book a single session or have a free trial. If they only sell 12-week prepaid packages, that's a sales tactic, not a teaching tactic.

## Where Pathwise sits

For context: each Pathwise session caps at 10 students, runs on real professional tools, and every learner walks out with a real, deployed web app. We sit in the realistic band for what hands-on, small-cohort AI classes taught by a working teacher cost to run sustainably.

We aren't the cheapest — and we don't try to be. If price is the only factor, there are cheaper options. If you want your child to walk out with something they actually built, this is roughly where the market is.

## So what should you do?

Three practical steps:

1. **Decide your format** first (online vs in-person, weekly vs camp). Each has its own price band; don't compare across.
2. **Compare like-for-like**. Three SGD 100 sessions of 10-kid in-person is not the same as three SGD 50 sessions of 30-kid online — even if the per-session price is similar.
3. **Always ask for a single-session trial** before a full term. Real programs offer this. Watch your child's eyes 30 minutes in. If they're building, the price is irrelevant. If they're staring at a slide deck, walk away regardless of price.

If you want to see what an in-person Pathwise session in Singapore looks like before booking, the [classes & schedule](/young-builders#schedule) page has the next two months of dates. Or [WhatsApp us](https://wa.me/6580261562) and we'll send you a recap from the most recent cohort.

— *Mr. Brown*


---

## AI classes for kids in Singapore: what to actually look for

> A parent's guide to AI classes for kids in Singapore — what they really teach, what to avoid, age-appropriate formats, and how much you should expect to pay.

Published: 2026-05-14 · updated 2026-05-25 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/ai-classes-for-kids-singapore

The best AI classes for kids in Singapore (ages 10–17) are **build-based, not theory-based**: every student should walk out with a real, deployed web app at a live URL. Look for small cohorts (~8 students per group), real professional tools (Claude, Figma, Midjourney), and an instructor who actually ships software — not someone reading slides. Online weekly cohorts typically run SGD 60–100/session, in-person workshops SGD 150–200/session, and school-holiday camps SGD 600–1,200 for the week. If a class ends in a certificate instead of a URL, skip it.


A parent stopped me at school pickup last week.

> "There are about fifteen new AI classes for kids in Singapore. I have no idea which ones are real. What am I actually looking for?"

It's a fair question. The landscape has gone from zero options to fifteen in roughly six months. Some are excellent. Some are coding camps with "AI" sprayed on the brochure. Some are a teacher reading slides about ethics while kids stare at the wall.

Here's the parent's guide I wish someone had handed me — what to look for, what to skip, and what to actually expect for a 10- to 17-year-old in Singapore in 2026.

## What "AI classes for kids" actually means in 2026

Three flavours dominate the market right now:

1. **Prompt-based classes** — kids learn to talk to ChatGPT/Claude. Useful, but it's the equivalent of teaching typing in 1998. Necessary, not sufficient.
2. **Build-based classes** — kids use AI to *make real things*. Web apps, designs, agents, working tools. This is where the actual skill lives.
3. **AI literacy / ethics classes** — discussions about bias, deepfakes, the future of work. Important, but a kid who's never *built* with AI is not equipped to think critically about it. Theory without practice is just opinion.

If you're choosing between the three, the build-based path wins by a wide margin. A kid who has made a working web app with AI has had a fundamentally different experience to one who has only talked to ChatGPT for homework help. They've discovered they have agency.

## What good AI classes for kids in Singapore look like

After running this kind of class for a few years, here's what I'd check:

### 1. Every student leaves with something live on the internet

This is the single most important filter. If a class promises "exposure to AI concepts" but doesn't end with the kid showing you a URL — a real web page on the actual internet that they made — it's almost certainly not worth your money. The shipped artifact is what proves the learning happened.

★ Test the brochure

Read the syllabus. If you can't find the words "deployed" or "live" or "URL" or "your kid's own app," ask the provider directly: *"What does my child walk out with on day one?"* If the answer is a certificate, walk away.

### 2. Real tools, not "kid versions"

Claude. ChatGPT. Midjourney. Figma. The same tools working professionals use every day. Kids handle real tools better than most adults — they don't have years of "this is too hard for me" baggage.

"Kid-safe" simplified versions teach a watered-down skill that doesn't transfer. They make parents feel safer; they make the learning worse. Look for classes that use the actual professional tools, with adult supervision rather than artificial guardrails.

### 3. Small cohorts (≤ 10 students, ideally)

You cannot ship a working app in a room of 30 students with one instructor. Full stop. The format requires real attention per kid — debugging their specific problem, looking at their specific design, giving feedback on their specific idea. Anything over ~10 per instructor becomes lecture-then-confusion.

### 4. The teacher should also be a builder

Ask: *"Is the instructor someone who has actually shipped software with AI, or someone who has read about it?"* Both can exist. Only one will give your child instincts.

In practice, this means looking at the instructor's portfolio. Have they launched apps? Have they built an AI agent that does real work? Do they have public projects? If yes, they'll teach in a way that compounds. If no, you'll get a syllabus and worksheets.

### 5. In-person matters more than you'd think for younger kids

For 10-12 year olds in Singapore, the in-person format wins. The energy of a roomful of kids problem-solving together, plus an instructor who can lean over and look at their screen, beats a Zoom call by a mile. Online formats start working well around age 14, when self-direction kicks in.

## Red flags to skip

In rough order of severity:

- **"AI literacy" or "AI awareness" with no building.** Theory-only classes treat your child as an audience, not a maker.
- **Massive class sizes** (20+). The format doesn't survive scale.
- **Worksheets, multiple-choice quizzes, "AI safety" modules** without any actual AI work. Often a sign of a school selling old curriculum with a new label.
- **No mention of what the kid takes home.** If the deliverable is "skills" rather than "their own working project," it's vapor.
- **Pricing under SGD 30 per session.** This is either a heavily subsidized intro, or the instructor isn't being paid enough to be good.

## What ages benefit most

A quick honest take after teaching hundreds:

- **Ages 10-11**: Sweet spot for short workshops and parent-present formats. Their imagination is huge and their willingness to try things is unmatched. They need a teacher in the room, not behind a screen.
- **Ages 12-14**: The strongest cohort. Old enough to self-direct, young enough not to be cynical about it. This is where I've seen the most dramatic builds.
- **Ages 15-17**: Can handle pro-level workshops at adult pace. The good ones will quietly outpace their parents within a few sessions. Treat them like junior developers, not students.

## What AI classes for kids cost in Singapore

Rough ranges I've seen for legit programs:

- **Online weekly cohorts**: SGD 60-100 per session
- **In-person Saturday workshops** (Singapore CBD venues): SGD 150-200 per session
- **Holiday camps** (full day, 4-5 days): SGD 600-1,200 for the week
- **Full school programs / hackathons**: scoped per school

You're paying for instructor quality, class size, and tools provided. The "good enough" threshold is around SGD 80 online or SGD 150 in-person. Below that, you're getting volume or volunteer instructors. Above SGD 250 per session, you're paying for prestige or one-on-one.

## Common questions

**Does my child need to know how to code first?**
No. The best AI classes for kids now teach building without traditional syntax — kids describe what they want, the AI writes the code, they direct and refine. The skill is *judgement* and *taste*, not memorising syntax.

**What will my child actually build?**
The good classes target a real, shippable artifact every session or every cohort: a deployed web app, a working AI agent, a designed brand system, or a small tool they use themselves. They should leave with a public URL or a downloadable file.

**Online or in-person — which is better for kids in Singapore?**
Under 14: in-person almost always wins. Over 14: either works, depending on the kid. In-person also gives you the side-benefit of meeting other AI-curious parents.

**Is this just glorified screen time?**
If the class is theory-heavy or chat-heavy, yes. If the class produces a shipped artifact every session, no — that's a craft skill being built, the same way drawing or music is.

**Can my younger child handle real AI tools?**
Yes, with an adult in the room for first sessions. By session three, most 10-12 year olds are running their own builds.

## Pathwise at a glance



## Where to go from here

Most parents I talk to start with one of two paths:

1. **Try it at home first.** I wrote a [step-by-step guide to introducing AI to your 10–17-year-old](/blog/how-to-introduce-ai-to-your-10-17-year-old) — three things you can do this Saturday before paying for anything.
2. **Join a structured class.** If you want guided instruction with small cohorts, we run [after-school and weekend classes for ages 10–17](/young-builders) in Singapore and online. Every learner leaves with a real, live web app and a URL they can show.

Whichever path you pick: look for the live URL at the end. That's the test that matters. Everything else is decoration.

— *Mr. Brown*


---

## How to introduce AI to your 10–17-year-old

> A practical guide for parents: four principles, three things to try this week, and one trap to avoid. From a teacher who's done this with hundreds of young builders.

Published: 2026-05-14 · updated 2026-05-15 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/how-to-introduce-ai-to-your-10-17-year-old

To introduce AI to a 10–17-year-old well: use **real professional tools** (Claude, ChatGPT, Figma), prioritise **building over chatting**, coach **taste** not just output, and set healthy habits from day one. Start this week with three at-home builds before paying for any class. When they're ready to go deeper — a real deployed web app, an AI agent — a small structured cohort (~8 students, every learner ships a live URL) is the next step. Pathwise runs these for ages 10–17 in Singapore, Hong Kong, and online.


Every parent I meet asks me a version of the same question.

> "I want my kid to be good at AI — but I also don't want them to become some kind of chatbot zombie who can't think for themselves. How do I do this right?"

It's a good question. And it has an answer.

After teaching AI to hundreds of young builders aged 10 to 17 — and raising my own — here are the four principles I keep coming back to, and the things you can actually try this week.

## Principle 1: Real tools, not toys

There is a temptation, especially with younger kids, to give them "kid-safe" simplified AI tools. I get it — the same instinct that hands a child plastic kitchen utensils before letting them near a real knife.

But here's the thing: **young builders can absolutely handle real tools.** Claude. ChatGPT. Midjourney. The tools working professionals use every day.

They handle them better than most adults, in fact. They don't have years of "this is too hard for me" baggage. They just open the thing and start poking.

The "kid-safe" versions teach a watered-down skill that doesn't transfer. The real tools teach the actual game.

★ Rule of thumb

If a professional would use it, it's appropriate for a 13-year-old — with you in the room.

## Principle 2: Build > Chat > Consume

There are three ways to use AI, ranked from most-valuable-to-learn to least:

1. **Build with it** — use AI as a tool to make a real thing (an app, a design, a video, a story they finish)
2. **Chat with it** — talk to it for answers or ideas
3. **Consume from it** — read what it produces, watch what it generates

Most kids encounter AI in reverse order. They first see AI-generated videos on social media (consume). Then maybe they try ChatGPT (chat). Almost no one starts at the top — *building* something.

But that's where the real skill is. And it's where the joy is too.

A young builder who has built a real working app, in a real coding tool, with their own idea, has had a fundamentally different experience to one who has only asked ChatGPT for homework help. They've discovered they have *agency*.

## Principle 3: Taste matters more than output

The AI will give you a million ideas. The skill — the skill that matters in the next 10 years — is knowing which ones are good.

This is harder than it sounds. We've all seen the kid (or adult) who generates 50 AI images and posts every single one. They've outsourced the *liking* of things to the AI too.

Teach your young builder to be picky. To say "no, not that — *this*." To know what they actually want before they ask. To throw nine out of ten things away.

> Taste is the muscle the AI age rewards. And it has to be exercised, not assumed.

The good news: kids develop taste fast when you ask them to defend their choices. *"Which version do you like best, and why?"* is the single best question I ask in class.

## Principle 4: Healthy habits, from day one

AI is going to be in their lives the way phones are in ours. Which means the habits formed at 12 will matter at 22.

Things I tell parents to set early:

- **Hard limits on "AI as homework helper."** Using AI to *understand* a topic is great. Using it to *write the essay* is a shortcut that costs them the actual learning. The difference is whether the kid could explain what they handed in.
- **Curiosity before consumption.** Before asking the AI a question, ask: *what do I think the answer is?* Then check.
- **A name for the feeling of "the AI did the work, not me."** In our classes we call it "ghost work." Kids spot it in each other instantly and call it out. Naming it is half the battle.

## 3 things to try this week

Here's a Saturday-morning version, ranked easiest to most ambitious:

### 1. Pick a design challenge together

Open Midjourney or any AI image generator. Pick a prompt together — *"a poster for a bakery on the moon"* or whatever. Generate 8 options. Have your young builder pick their favourite and *explain why*. Then have them write a new prompt to improve it.

15 minutes. They'll have opinions. That's the point.

### 2. Build a one-page tool

Open Claude or any AI coding tool. Pick something useful — *"a website where I can paste in a song and it tells me the year it came out"* or *"a homework timer that plays a sound every 25 minutes."* Build it together. Get a real URL at the end.

45 minutes. They've now made software. That doesn't go away.

### 3. Have the "what's the AI doing?" conversation

Watch them use an AI for 10 minutes. Then ask:

- "What did you actually do that the AI couldn't have done?"
- "What did the AI do that you couldn't have done?"
- "If you had to do this without the AI, what would change?"

This is the conversation that builds the meta-skill. Don't lecture. Just ask.

## When a guided class makes sense

You can do a lot of this at home. But the moment your young builder needs to go deeper — actually build a real, deployed web app, learn to direct a design tool with intent, or build their first AI agent — that's where a structured cohort wins.

We run [after-school and weekend classes for ages 10–17](/young-builders) where every young builder walks out with a real, live web app and a URL they can show. Small cohorts (8 students max), real tools, taught by working teachers who actually build with this stuff.

If that's where you're heading, [come say hi](/contact). If not, the four principles above will get you a long way at the kitchen table.



The kids will be fine. They just need someone showing them the door.

— *Mr. Brown*


---

## How we run an AI hackathon at a school (anatomy of a build day)

> 200 students all building at once isn't chaos — it's the most focused energy you'll see in a school all year. Here's exactly what a Pathwise hackathon day looks like, what schools need to provide, and the questions every head of school asks before booking.

Published: 2026-05-14 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/how-we-run-an-ai-hackathon-at-a-school

The first time a head of school asks if we can run a full-school AI hackathon, there's always a pause.

"Two hundred students, all building at the same time, in our gym?" they ask. "Isn't that going to be chaos?"

It's a fair question. And the answer is no — but not because we're miracle-workers. It's because a well-designed hackathon day has a rhythm that *makes* focus inevitable. By 10:30 AM the energy in the room shifts, and from then on the only sound is keyboards and quiet conversations between teammates.

Here's what actually happens.

## A hackathon day, hour by hour

This is what a typical full-school Pathwise AI hackathon looks like at a primary or secondary school. Day one of a one- or two-day event. Times shift a bit by school schedule, but the shape is consistent.

### 8:00 AM — Setup

We arrive an hour before students. Tables are arranged in pods of 4–6 (we coordinate this with your facilities team a week ahead). Each pod has a small theme card and one of our mentors floating to a cluster of pods. Wifi tested. Coffee on. Sound system checked.

### 9:00 AM — Opening assembly (30 minutes)

Whole school in the hall or theatre. Short, high-energy opening from one of our lead instructors. Not a lecture — a live demo. We build a tiny working app in 8 minutes in front of everyone. The room goes from polite-attention to *"wait, can I do that?"* in under five minutes. That's the unlock.

We then explain the theme of the day (more on that below), reveal the prize categories, and send the students to their pods.

### 9:30 AM — First build session (90 minutes)

Pods get a structured prompt sheet (we provide). It walks them through: pick an idea, sketch it, start building. Mentors circulate continuously, unblocking. Every 20 minutes we ring a soft bell and check in — *"raise your hand if you've shown your idea to someone outside your pod. Now go do it."*

By the 60-minute mark, every pod has a half-working version of something. Some look amazing. Some are still finding their footing. Both are fine — this is the part where we want raw experimentation, not polish.

### 11:00 AM — Quick demos (30 minutes)

Each pod stands up and gives a 60-second show-and-tell of where they are. The point isn't to compete yet. The point is *exposure to other people's ideas* — most pods leave this session pivoting or expanding what they're building, having seen what's possible.

### 11:30 AM — Break (30 minutes)

Snacks. Movement. Outside if there's a courtyard. We deliberately don't keep them inside the whole day — energy comes back stronger after a real break.

### 12:00 PM — Build session two (90 minutes)

The longest unbroken work block of the day. Now they know what they're building, who their teammates are, and what's possible. We see real apps starting to come together. Mentors shift from "unblocking" to "pushing for polish." If a pod is ahead, we ask harder questions. If a pod is behind, we help them re-scope to something they *can* ship.

### 1:30 PM — Lunch

Provided by school as agreed in the brief.

### 2:00 PM — Final build + deploy (75 minutes)

The deploy session. Every pod gets their app on the internet, with a real URL. This is the bit students remember forever. *"Wait — my Auntie in Canada can actually open this on her phone right now?"* Yes. She can. Send her the link.

### 3:15 PM — Showcase (45 minutes)

Closing assembly. We pick 6–10 standout pods to present to the whole school (we identify them during the day). Each presents in 90 seconds. We project their app live. The room responds.

Then prizes — usually 3–5 categories: most-creative, most-useful, best-collaboration, judges'-favourite, audience-favourite. Every pod that presented gets recognised in some way.

### 4:00 PM — End of day

The hall empties. Phones come out. Group photos. Students texting URLs to parents. Teachers wandering up to ask *"how did you get them to focus like that for six hours?"*

We pack down. We're out by 5.

## What the school provides vs what we bring

This is the question every head of school asks, so here's the clean version:


  **Schools provide:**

  - A room with WiFi (gym, hall, theatre, large classroom — we've worked with all)
  - Student laptops (or guidance — we can advise on rentals if needed)
  - 1 lead teacher liaison + classroom-teacher coverage during sessions
  - Lunch + snacks for students
  - Any specific theme or curriculum tie-in you want to incorporate

  **We bring:**

  - Full hackathon curriculum and prompt sheets, age-appropriate
  - 4–8 mentors (ratio scales with cohort size)
  - AI tooling accounts pre-configured for every student
  - Hosting / deployment infrastructure
  - Theme cards, prize structure, opening + closing scripts
  - Photo + highlight video of the day (optional add-on, popular for parent comms)


## The worries that come up (and the real answers)

Heads of school, in my experience, ask versions of the same five questions. Let me answer them straight.

### "Our teachers don't know AI well. Will they be embarrassed?"

No — but only because of how we run it. Teachers aren't expected to lead anything technical. Our mentors do that. The teachers' role is what they're already great at: pastoral, watching for kids who need a check-in, helping with classroom flow. Many tell us afterward they learned more about AI by walking the room than they would have in a teacher-training day.

### "What if a student gets stuck and feels left out?"

This is the worry parents ask too. Our pod structure is designed for it — every pod has at least one student who's faster on the keyboard and at least one who's quicker with ideas. Mentors are watching for the disengaged kid and re-engaging them with a job (the "designer" role, the "tester" role, the "demo presenter") that suits how they actually contribute.

It's rare for a student to fully check out on a hackathon day. The energy in the room is too high. But when it happens, we catch it within 10 minutes.

### "We're nervous about what they might *make*"

A legitimate concern. We use AI tools with strong content guardrails and we monitor outputs throughout the day. Inappropriate content is essentially never an issue — and when something edge-case comes up (a 13-year-old asking the AI something silly), our mentors catch it and have a quick teaching moment with that pod.

We can also share our approach to safe AI tool usage with your safeguarding lead ahead of time. Many schools want this documented.

### "How do we tie this to our curriculum?"

We theme the day around your school's priorities. If you're a sustainability-focused school, the theme is "build an app that helps the planet." If you're STEAM-heavy, "build a tool that solves a problem you've noticed in school." If you're tying it to a year-long inquiry, we map the theme to it. The students get a structured prompt that funnels the creativity into something on-curriculum.

### "How long does it take to organise?"

A clean hackathon day takes about 6–8 weeks from signed brief to event day. That covers theme scoping, mentor briefing, tooling provisioning, photo permissions, classroom prep, and a parent-comms package. Tighter timelines are possible but tighter than 4 weeks gets uncomfortable.

## Outcomes (the bit that matters)

A few things you can expect by the end of a Pathwise hackathon day:

- **Every student has built something real.** Not a worksheet. Not a poster. A live, working web app on the internet.
- **Every student has a URL they can show parents, grandparents, friends.** This is the bit that makes it feel real to families. Many parents tell us afterward this was the first time their kid had created something on the internet that *worked*.
- **The school has stories.** The press-ready recap (photos, demos, a couple of pod highlights) is something you can put in newsletters, on Open Day, and in your next round of admissions communications.
- **Teachers leave more curious than they came in.** This is the underrated outcome. A successful hackathon shifts how teachers see AI — from "the thing students are using to cheat" to "the thing students are using to build."

## If you're thinking about it

We run hackathons at schools across APAC — single-year-group events, full-school days, multi-day intensives. Primary and secondary, with curriculum tailored per age.

The fastest way to see if it's right for your school is a [free 30-minute discovery call](/schools#book-discovery). We'll ask about your priorities, you'll ask whatever you want, and we'll send a written proposal a few days later.

Or [browse our school programs](/schools) for the full menu of options — weekly afterschool, holiday camps, residencies, and full-school hackathons.

The students will surprise you. They always do.

— *Mr. Brown*


---

## What is vibe coding? A 4-minute primer for non-coders

> Vibe coding is the fast track from idea to live web app — no syntax required. Here's what it is, what it isn't, and how to start today.

Published: 2026-05-14 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/what-is-vibe-coding

A friend asked me at dinner what "vibe coding" actually means.

She'd seen the term on LinkedIn, watched a video where someone built an app "without writing any code," and walked away convinced one of two things had to be true: either it was a trick, or every developer she knew was about to lose their job.

It's neither. Here's the honest version.

## The 10-second definition

**Vibe coding is software development where a human describes what they want in plain English, and an AI agent writes, runs, and fixes the actual code.** You're directing. The AI is doing.

You still need taste. You still need to know what you want. You still have to read the output and say "no, not like that — like *this*." But you don't need to remember the syntax for a `for` loop or how to deploy to a server. The AI handles that part.

That's the whole thing.

## How it actually works

A vibe-coding session, in 2026, looks something like this. You open a tool like Claude Code (the one I use). You type:


  *"I want a simple web app where I can paste in a recipe URL and it tells me how many calories per serving. Make it pretty."*


The AI thinks for a few seconds, writes ~200 lines of code across a handful of files, runs it, and shows you a working website. You look at it and say:

> "The colors are ugly. Can the calorie number be huge and the rest small?"

The AI rewrites the styling. You refresh. The number is huge. You say:

> "Now deploy this to the internet so I can share it with friends."

The AI does. You get a URL. You text it to three people. They open it on their phones. It works.

The whole thing took 25 minutes.

## What it isn't

Three things that vibe coding **isn't**, because every other article I've read seems to get one of these wrong:

★ It isn't magic

The AI will write code that looks like it works but is subtly broken. You still have to test it. You still have to ask "wait, what happens if a user does *this*?" The AI will not catch every edge case, and if you ship without thinking, you'll ship something embarrassing.

★ It isn't "AI does it all"

Taste matters more, not less. The AI will give you a working app. It won't give you a *good* app. Good apps come from someone who knows what "good" means — which is you. The skill being rewarded here is judgement, not typing speed.

★ It isn't the end of "real" coding

People who already know how to code are getting 5–10× more productive with these tools, not unemployed. What's changing is the bottom of the on-ramp: someone with zero coding background can now build something real, fast. That's new.

## Who it's for

Honestly? Anyone with an idea.

- The founder who keeps saying "if only I could build a prototype myself"
- The marketer who wants a working version of the campaign tool they're describing to engineering
- The designer who's tired of mocking things in Figma and wants them clickable
- The parent who wants to build their kid a custom homework helper that doesn't just give answers
- The teacher who keeps thinking *if only this lesson had an interactive piece*

If you've ever finished a sentence with "...but I don't know how to code" — vibe coding is for you.

## A 1-hour starter exercise

If you want to try this today, here's what I'd do:

1. **Sign up for [claude.com](https://claude.com)** (free tier is fine to start). Or any equivalent AI coding tool.
2. **Pick one specific, small idea**. Not "an app for my whole business." More like: "a one-page web tool that does *one specific thing* I keep doing manually." A unit converter. A meeting-notes summariser. A pet-name generator. Make it small.
3. **Describe it as if you were briefing a really fast intern**. What does the user see first? What do they click? What do they get back? Be specific.
4. **When the first version is bad, say what's bad**. Don't rewrite from scratch — iterate. "The font is too small." "The button should say 'Convert' not 'Submit'."
5. **Deploy it**. Most tools have a one-click deploy. Get a URL. Send it to one human you trust.

Total time: ~1 hour, give or take, if you've never done this before.

## What happens next

You'll either feel a giddy "wait, *I* did this?" or a flat "this isn't for me." Both are useful answers. But the giddy version is more common than people expect — and once that switch flips, you don't go back.

We run live four-hour workshops where we walk adults through exactly this — idea to deployed app, in one Saturday morning. If you'd rather do it with a room full of people, coffee, and someone to ask "wait, why did it do that?" — [browse the next dates](/adults#schedule) or [drop us a line](/contact).

The hardest part of vibe coding isn't the coding. It's giving yourself permission to try.


---

## What real AI training for teams actually looks like

> Most corporate AI training is a slide deck nobody remembers. Here's what we do instead — and why every workshop ends with something built, not a quiz.

Published: 2026-05-14 · Author: Alan Brown · URL: https://pathwiseacademy.co/blog/what-real-ai-training-for-teams-looks-like

A head of marketing called me last month, two weeks after their team had finished a "company-wide AI training programme" with a big consultancy.

"They watched four hours of videos," she said. "There was a quiz at the end. Everyone passed. And now nobody's actually doing anything differently."

She's not alone. Most "AI training" I see being sold to teams in 2026 is the same training people were getting in 2023, just with newer screenshots. Slides, talking heads, a "framework," a quiz. Then the team gets back to work and… nothing changes.

Here's why, and what works instead.

## Why slide-deck AI training doesn't transfer

Three problems, all of them obvious in hindsight.

★ Problem 1

**Watching is not doing.** Nobody learns to ride a bike from a deck on bike-riding. AI is the same. You don't *understand* prompt engineering until you've sat with a real tool, given it real input, gotten frustrated, fixed it, and shipped something. Slide decks bypass the muscle the team actually needs to build.

★ Problem 2

**Generic examples don't transfer.** "Here's how AI could help with sales emails" lands as theory if the example isn't your team's sales emails. People nod, take notes, and never act on it because the example was abstract and the bridge to their actual job was left unbuilt.

★ Problem 3

**No artefact = no momentum.** A team leaves the training with… what? Notes? A certificate? Nothing they can show their colleagues, nothing they can keep iterating on, nothing to point at next Monday and say *"we did this in the workshop."* The behaviour dies on the bus home.

## What we do instead

Every Pathwise corporate workshop is built around one principle: **your team ships something real before they leave.**

A typical four-hour Pathwise corporate session looks like this:

- **0:00 — Coffee, intro, and "what are we actually trying to build today?"** Not generic AI literacy. A real, specific outcome someone on the team cares about. We pre-scope this with the team lead before the day so the room walks in pointed at something.
- **0:30 — Live demo of the toolchain on their problem.** Not "this is how Claude works in general." This is "let's open Claude and start solving the thing you brought in." Five minutes in, the room is already engaged because the example is theirs.
- **1:00 — Hands-on building, in small groups.** Pairs or trios. Each pair takes a piece of the team's actual workflow and builds an AI-augmented version of it. We circulate, unblock, push back, suggest. Nobody is watching slides.
- **3:00 — Demos to the room.** Every group shows what they built. Real artefacts: prompts, working tools, deployed apps, drafted copy. Five minutes each.
- **3:45 — "What you're taking back."** A 15-minute structured conversation about which of these get adopted, who owns each, and what the next two weeks look like.

By 4:01 PM, the team has built five real things, knows how to repeat the process, and has a written list of what comes next.

That's it. There's no magic. The magic is that every minute was spent doing the actual thing, on the actual problem.

## A real example

Last year a marketing team came to us with a recurring pain: they were spending 8–10 hours every Monday compiling the week's campaign performance into a Slack-ready summary. They wanted that down to under an hour.

We didn't teach them "AI for marketing analytics." We sat down with their actual dashboards, their actual Slack format, their actual brand voice, and built a working Claude-driven workflow that pulled the raw numbers, summarised what changed week-over-week, drafted the Slack post in their tone, and flagged outliers for the team to review.

By the end of the day they had:
- A working prompt template they ran every Monday
- A documented hand-off from "AI does first draft" to "human reviews and posts"
- An estimate of ~7 hours/week saved across the team

We didn't tell them they'd save 7 hours. They figured it out themselves, on the whiteboard, after building the thing. That's how it sticks.

## The ROI conversation

People ask me about ROI on AI training. Honestly, I think it's the wrong question — or at least the wrong framing.

The right framing isn't *"how many dollars per training hour do we save?"* It's *"how much faster can your team move on the things you already want to do?"* AI training that works doesn't replace people's jobs; it removes the friction between their idea and their output.

The unlocked velocity is the ROI. It's just hard to measure in a quarterly review.


  A practical test: six weeks after the workshop, ask the team "what's something you do differently now?" If they can name something specific without thinking, the training worked. If they say "uh… we talk about AI more, I think?" it didn't.


## What this means for you

If your team is about to do AI training — your own or vendor-led — push on these questions:

1. **What will my team actually build during this session?** If the answer is "nothing, but they'll learn," walk away.
2. **What does my team take home that they can use Monday morning?** A deck doesn't count. A working artefact does.
3. **Will the examples use *our* problems, or generic ones?** Generic examples are a red flag. Either you're paying for off-the-shelf content, or the trainer doesn't know how to adapt.
4. **What does success look like 6 weeks later?** If the trainer doesn't have a clear answer, they're not building toward retention.

We run corporate AI workshops across Singapore, Hong Kong, and APAC — half-day, full-day, weekend formats. Every session ends with something real your team has built. Onsite at your office or online for distributed teams.

If you're scoping a workshop, [book a free 30-minute discovery call](/companies#book-discovery) and we'll talk through the specific outcomes you want. No slides on that call either.

— *Mr. Brown*