Bring an idea, or something you’re tired of doing over and over. You leave not only with a webpage, tool, application or workflow you’ve “talked” into being, but also with a way to keep turning future ideas into something that works.
Who teaches
Horris Tsebio ↗
Finance · Technology · AI · Product · Communication
20+ years in programming & product development
This session is full. If a seat opens (unpaid holds sometimes lapse), the waitlist is offered it in order — leave your email to join.
FPS · PayMe, no fee. Stripe / PayPal +3%. Registering creates your HtPi account — members save on every class. Transfer & refund rules → · Gift a seat →
Vibe coding lets you describe what you want in natural language, and have AI help you build, change and test webpages, small tools, simple apps and workflows.
You don't need to know programming to start.
This class walks you through the full journey once — from an idea to a version that works.
You get a first version — then one change and it all starts to wobble?
Tried a few tools, still not sure which one you should be using?
AI says it's done — but you can't actually tell whether it is?
Stuck, so you prompt again — and every round drifts further?
That is exactly what this class works on: breaking down, changing, testing, judging — and finding the problem again when something goes wrong.
Bring your idea.
We'll narrow it to a version that fits in three and a half hours, then build the first working version together.
You don't need to start out imagining a whole piece of software.
About an hour and a half of the class is hands-on. With a clear plan, a small web app with sign-in, a database and an AI feature is within reach.
Customers pick a real time slot and book it; you manage every booking from an admin page. Not a page with a form on it — the bookings are stored, and each one gets a confirmation email.
Upload photos or PDFs of receipts and invoices. AI reads the shop, date, amount and category into a table you check and correct, then it totals them by month and exports them.
Paste in orders the way customers actually send them. AI turns each one into a structured order; the board follows it from new to packed to delivered, keeps each customer's history, and gives you a daily summary.
Upload a handbook, price list or policy. Staff or customers ask in plain words; the answer quotes the page it came from, and anything it can't answer is logged for you to fill in.
The host opens a room; players join with a code on their phones; each question appears for everyone at once, and the leaderboard updates live. AI drafts the questions from your topic or notes — you choose which to keep.
The goal of the three and a half hours is not to finish a large product. It is to take one real, well-scoped problem from an idea to a usable first version.
Don't be put off by the dazzling one-prompt apps online. Many aren't built around a real need, and some look finished but do nothing when you press the buttons. Making a website is easy now anyway — the hard part is making something you'll actually use, that actually works. That's what the class practises.
The examples this class series used in its earlier runs, kept here for reference. Models improve fast, so examples date quickly — but the skill of building something useful stays the same.
Not just a static introduction page. You can add content, schedules, forms, and a payment flow.
Turn repetitive Excel or CSV work into a small tool: upload the data, clean it automatically, then download the result.
Turn scattered, inconsistent inputs into a fixed structure, then save them automatically or write them into a spreadsheet.
Build a simple app around one specific problem, with a clear input and result — for example, calculating, searching, organising, or processing content.
Build a game with straightforward rules — for yourself, to share with friends, or as a simple practice activity.
Getting AI to put a screen together is no longer hard. Where people actually get stuck:
“I want to build an AI app.”
Version A works. Then you ask:
“Add a filter.”
The whole layout shifts. Things that worked stop working. You are further away than before you asked.
Learning to make the second path your default is a big part of this class.
This class doesn't teach the buttons of one particular tool — it teaches you to run the whole process.
This class doesn't bet on any one vibe-coding tool.
Tools change, but a few questions stay: how to take an idea apart, how to pick the right tool, how to keep AI on course through changes, how to test the result — and how to find the problem again when something breaks.
That process is what we learn.
HtPi.me and MasterPokerApp.com are both projects I built — and keep building — with an AI-assisted, vibe-coding workflow.
They are, of course, far beyond the scope of a three-and-a-half-hour class.
What the class does is teach you those first steps: how to take an idea apart, make a first version, and keep changing it with a method.
Horris Tse
bio ↗
I don't teach vibe coding as a piece of software. It has become part of how I actually build products, websites and workflows.
HtPi.me, MasterPokerApp.com, and a good number of day-to-day prototypes were all built step by step this way.
The class takes that method apart — including the holes I've stepped in along the way.
Vibe coding is a new way of making things: you describe what you want in natural language, and AI helps you build, change and test webpages, small tools, simple applications — even everyday workflows.
That lowers the barrier to "making the thing" a great deal. Even without a programming background, you can start from one sentence and gradually make something genuinely usable.
But getting AI to write code is not the same as knowing how to build something well.
Once requirements blur, features start to interact, errors appear, or data, APIs, accounts and security are involved, you still need to know how to break the problem down, judge AI's answers, and know when to stop and rethink.
The class uses whichever AI coding tools currently fit. Budget around HKD 300–500 for tools — e.g. a VPN and an AI subscription, paid directly to those providers, not to HtPi. Exactly what's needed is listed clearly before you register — you will never be asked mid-class to subscribe to a pile of products.
Yes. The class starts from breaking an idea down and assumes no coding background; you'll pick up enough of the underlying structure to know what you are asking AI to do.
We demonstrate with tools that currently fit the job — but the point of the class is the process itself. Tools change; the method doesn't.
A laptop (and its charger). We set the tools up together in the first part of the class.
A small problem you genuinely want solved — organising some data, a page with one purpose, a repetitive task. The thing you build does not need to be complicated.
We narrow it together into a first version that fits in three and a half hours. Big ideas are exactly what small versions are for.
Please budget around HKD 300–500 for tools — e.g. a VPN and an AI subscription — paid directly to those providers, not to HtPi.
Full policies →Horris is highly experienced in the field. He presents complicated and abstract AI concepts in a simple, accessible way, using plenty of live examples. With his guidance, I've already started exploring different business ideas using the agent I built, and I highly recommend this course.— Michelle W. · Multi-Family Officeattended: OpenClaw Hands-On Workshop · 20 Apr 2026Reserve a seat
I use Google analytics and ad-measurement cookies — on by default — to see what gets read and whether ads work. Decline removes them, leaving only cookie-free, non-identifying statistics. See the Privacy Policy. I use Google analytics and ad measurement. Cookies are set only if you accept — until then Google receives only cookie-free, non-identifying statistics. See the Privacy Policy.
