Weekend Startup Lab: building a real product with AI

Most of us have seen the demo. Someone types a single prompt, and a few minutes later an app is on screen, styled and deployed. It looks as if the hard part of building software has been automated away.

Then the real product arrives. Customers do unexpected things, data turns up messy, and accounts, passwords and payments carry real consequences when something breaks. AI writes most of the code, and the last stretch, the part that decides whether a product ships, still needs a person.

Weekend Startup Lab is built around that last stretch. It is a two-weekend, hands-on workshop that teaches people with no coding background how to build a real product with AI, and how to keep building after the course ends.

How the two weekends work

The course runs over two in-person weekends, Saturday and Sunday each, with a week in between. That gives four days in the room, plus one online office-hours evening during the gap for questions about whatever you are building.

We keep the framing short. At most one day goes on the problem and the people you are solving it for, so most of the time goes on building. You have a few light conversations with potential customers before the first weekend, and deeper ones between the two weekends, once you know what you are making.

The thinking behind this is Humanistic Thinking™, the framework Acoustic uses on client work: listen to real people, make sense of what you heard, decide what matters, then build it and test it. Erik Ingvoldstad leads every cohort, working with a local co-facilitator in each city.

Weekend one sets you up to build. You settle on the problem and the person you are solving it for, turn that into a short plan, set up your accounts and tools, and get a first version running.

Weekend two is about finishing. You bring back what you learned from customers, change the product to match, and work through testing, security and privacy so it can face real users.

What you learn

We teach a working model with three roles. Claude is the thinking partner, talking through the idea, the architecture and the trade-offs before anything gets built. Claude Code is the pair of hands, writing and editing the code inside your project. You supply the judgement: you decide what is right for the product, test it for real, and own whatever ships.

AI does most of the typing, so our time goes on the habits that keep a product standing once real people use it:

• Planning first. Turning a brief into a build plan, and writing a CLAUDE.md file with your stack and rules before your first prompt.
• A repeatable stack. The same handful of tools on every project, such as VS Code, Next.js, Firebase, Sanity, GitHub and Vercel, with Expo for mobile apps. Which ones you use depends on what you are building.
• Keeping secrets secret. How to keep keys and passwords out of your code and out of your chats.
• Shipping. How a change moves from Claude Code to GitHub to Vercel and onto your live site automatically.
• Testing. Trying your product with messy data, and with someone who is not you.
• Security and privacy. Sanitising user input, reviewing authentication rules, knowing what personal data you store, and what Singapore's PDPA or GDPR expect of you.

You leave with a working product built from your own idea, and a clear route to getting it in front of users. You also leave with a setup you can reuse: the accounts, tools, project instructions and deployment pipeline we use on our own products. Your next product, or the next version of this one, starts from there.

Just as useful is knowing what to check before real users arrive. Most people who build with AI hit a wall once data gets messy or accounts and payments come into play. You meet those problems here, with a facilitator beside you.

We have already run a first session outside the Weekend Startup Lab name, and the response has been encouraging. One participant described it as a fantastic learning experience for the whole group.ou leave with a working product built from your own idea, and a clear route to getting it in front of users. You also leave with a setup you can reuse: the accounts, tools, project instructions and deployment pipeline we use on our own products. Your next product, or the next version of this one, starts from there.

Just as useful is knowing what to check before real users arrive. Most people who build with AI hit a wall once data gets messy or accounts and payments come into play. You meet those problems here, with a facilitator beside you.

We have already run a first session outside the Weekend Startup Lab name, and the response has been encouraging. One participant described it as a fantastic learning experience for the whole group and recommended it to risk managers across Asia and beyond.

Who it suits and where it runs

No coding experience is needed. Aspiring and current founders, business executives, students and anyone with an idea and no technical co-founder have a place here, as do people who simply want to understand how AI-built products come together. You need a laptop, the ability to use a computer and web services, and something you want to build.

Singapore comes first, on 28-29 November and 5-6 December 2026. The course then travels to Dubai, Kuala Lumpur, Bangkok, Manila, Hong Kong, a second Singapore cohort, Munich, Zurich, Oslo, Sydney and Melbourne, through to October 2027. Dates, pricing and registration are at weekendstartuplab.com, and questions go to hello@weekendstartuplab.com.