Try it,
rather than take my word.
First fifteen years of projects, then thirteen demos to try, picked because they make sense in a minute. It isn't a catalog: at your company, I build what the way you work calls for.
What I've already built
From startups to large companies. Most of it never had a public page, let alone a demo.
A whole platform, from scratch
Co-founder and CTO: web and mobile apps, automated recruiting, a team of 20 developers and nearly 90,000 temp workers on the platform.
An internal tool for a large company
Managing the wholesale of international phone minutes.
The software of a vending machine
The application for France's first stamp vending machine.
Image recognition
"Color Your Summer", a web game that recognised coloured cans.
Community platforms
A neighbourhood social network in New Zealand, a platform for independent musicians in Japan.
Online courses and enrolment
A course platform and enrolment flow built during Covid, so classes could carry on remotely.
Real time
Live interactions in Node.js, and elephant.io, an open-source library to talk to websockets from PHP.
AI agents under human control
A platform where AI agents prepare the work and humans decide.
Thirteen demos to try
On a phone, the demos are shown as images, step by step. To try them for real, with live AI on your own data, open this page on a computer.
- 1WatchMost have a one-minute guided tour, on fictional data. Next, Previous or the arrow keys.
- 2TryTake over: click anything, or bring your own message, document or question.
- 3Let's talkThe button at the end sends me what you tried. 20 minutes to see what it would look like at your company, no commitment.
This badge marks what genuinely runs: what you type goes to Claude, Anthropic's model, and the answer is written in front of you. The tour itself is written in advance. Each call costs money, hence a few tries a day. Don't put anything confidential in it.
On your data, right now
No script: they run on what you give them, your website or your need, with a real model.
Client requests, quotes and follow-ups
What comes in from your clients and goes back out: the AI reads, understands and prepares, you approve.
Documents and data
PDFs, scans and messy spreadsheets, made machine-readable, then genuinely useful.
The team, control and where to start
Day-to-day use, what the AI may do on its own (nothing irreversible), and where to start.
Code and tech teams
For founders who built a tool with a coding assistant, and teams who want to use one without damage.
Built your app with AI and have real customers? Taking it to production →Your case isn't here?
That's the most common case. These demos are examples: at your company, I start from your tools and the way you work.