AI EXPERTISE

I don't talk about AI.
I ship with it.

For over a year I've been PM on an AI-native product, earned a Gen-AI product certification, and rewired my daily craft around AI tools. This is what working in AI, not around it, actually looks like.

ask · analytics
01 · BUILD IT

A year inside an AI-native product.

I spent nearly a year as PM on Toucan AI, a conversational analytics product. The biggest lesson: building an AI product is nothing like building a regular one. The real design work happens as much in the model's behavior as in the interface.

It also means accepting that the product is probabilistic by nature, and designing for it: building trust, setting the right expectations, and making the AI's reasoning legible to non-technical business users.

EXPERIENCE
Conversational UX
QUALITY
LLM Evaluations
ARCHITECTURE
Multi-agent systems
SOVEREIGN INFRASTRUCTURE
Mistral & Scaleway
Read the Toucan AI case study →
02 · WORK WITH IT

AI, distilled into how I ship.

I use AI to move faster and deliver better, every day. It reshapes how I work with engineers and how quality gets shipped.

Living specs

Product requirements that live inside the codebase and can be queried. Specs that stay true to what actually shipped.

Vibe coding

UI/UX iterations straight in Cursor and GitHub, shipping small improvements myself instead of queueing dev time.

Agents in production

Live agents doing real work: user-feedback analysis, competitive monitoring, and product documentation. n8n in 2025, Claude agents in 2026.

03 · PROVE IT

I ship with AI. Here's the receipt.

My GitHub contributions are the simplest proof that I've folded AI into my daily work, whether I'm writing specs or vibe coding.

Contribution activity · 2026
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Mon
Wed
Fri
View on GitHub →
Less More
POINT OF VIEW
Most AI features fail not because the model is wrong, but because the product never makes it obvious when to trust it and when not to. Building that trust layer is the real PM job in AI.
GROUNDED IN THEORY
GenAI · Les Fondamentaux, AI Discipline · Tristan Charvillat
CERTIFIED · 2025

GenAI · Les Fondamentaux

The GenAI foundations program by AI Discipline, led by Rémi Guyot (ex-Booking.com CPO) and Tristan Charvillat (VP Design, Malt). Theory to back up a year of hands-on practice.

Prompting methodology
RAG & knowledge-base structuring
Building & configuring AI assistants
View the certification →
TOOLS & STACK

The kit I build with.

A modern, connected stack, wired together so the tools talk to each other instead of sitting in silos.

AI & LLMs
Claude Claude logo
ChatGPT ChatGPT logo
Mistral Mistral logo
Build & specs
Cursor Cursor logo
GitHub GitHub logo
Lovable Lovable logo
Figma Make Figma Make logo
Claude Design Claude Design logo
Analytics & evals
PostHog PostHog logo
Langsmith Langsmith logo
Custom eval tooling
Automation & agents
Dust Dust logo
n8n n8n logo
Claude API Claude API logo
MCP MCP logo