Coffee Chat
We talk. Someone shares the thing that's eating their week, everyone else listens and throws out ideas.
The clever machines are ready. The real question is what they're for.
Most of the AI world treats a human in the loop as a box to tick. We think the human is the whole point. So we made a small room where the people doing social work and the people who know the tools sit at the same table — and the technology serves the mission, not the other way around.
Who's in the room
Half of us do the work. Half of us build the tools. We've just never been in the same room.
You spend your days with people, not products.
You already know exactly where the week leaks — the third form, the report nobody reads, the funder who wants numbers by Friday. You don't need a lecture. You need someone who'll build the fix while you point at the problem.
You spend your days with the tools.
You know exactly what they can do. You've just spent three years aiming that at getting people to click ads. Here's a room full of problems that are actually worth your Saturday.
What makes this different
Not another AI course. Not a consultancy. A room with a purpose.
The people who do the work lead.
The sector is full of AI training sold to NGOs by people who've never done the work. Here, the person with the problem decides what's worth fixing. The tech folks listen, then build to that — not the other way around.
Built for Hong Kong's social sector, not copied from Silicon Valley.
Our skills library is grounded in the standards your funder actually asks about — SWD reporting, LSG service contracts, PDPO data rules — not a generic startup playbook bolted onto a charity.
Free, open, and nobody's sales channel.
No upsell waiting behind the door, no kickbacks, no selling your data. Everything we make is open for anyone to take and reuse. Not now, not later.
One thing that works by Monday.
We don't sell a platform or a six-month roadmap. We fix the one task eating your week, hand it over with instructions, and leave. If your team can't run it themselves, we haven't finished helping.
Three ways to show up
Three ways to show up. All free.
Saturday Project
We build. Pick one real problem, fix it by lunch, hand it over with instructions.
Agent Onboarding
We set up your first AI agent with you — the install, the models, the skills — then keep the tips flowing.
Why now
The door's open now. It won't stay this wide.
The first meet-up is July 2026. Come now and you're in the founding cohort — the people who decide what this room becomes before it has a reputation to lean on.
Small on purpose
20 to 30 people. We read every application and turn away the ones who came to sell. That won't scale forever.
The window is now
AI is being pitched to your sector hard, by people who don't know the work. This is the moment to set the defaults yourself — with people who do.
Free while it lasts
Volunteer-run, so it's free today. The skills library exists now, not on a roadmap. The sooner you're in, the more it bends to your week.
The door
Bring the one task that ruins your week.
The data mess. The report no one reads. The thing you're quietly embarrassed you still do by hand at 9pm.
Tell us that — plainly — and you're in. The founding cohort shapes this room. Come early, and it bends to your mission instead of the other way around.
20–30 people. An independent café, after hours. We read every application — and we say no to the ones who came to sell. Where and when, once you're through the door.