AI Product Manager. Turning AI agents into products that generate revenue: observable, evaluated, auditable
Most AI pilots never prove their revenue impact. I ship the ones that do and prove it in euros.
10+ years turning messy, regulated data into products that move the number.
I build and ship products with impact, not decks:
- +30M€ unlocked through data quality
- +8M€ from a pricing product at Decathlon
- 300M€/year billing platform at BNP Paribas
- −75% production time on core reporting
Built observability platforms tracking AI adoption across the BNP Paribas Group. Run data teams, roadmaps, and stakeholders in regulated industries where failure costs millions.
Then I started building for myself.
| Project | What it does | Status |
|---|---|---|
| SASU IS Simple | Salary vs dividends optimizer for French freelancers | 🟢 Live |
| The AI Compass | 10-day email course — AI for non-technical professionals | 🟢 Live |
| EvalBlink | Benchmark prompts × models × tests — quality, cost, latency in one command | 🟢 Live |
| ValuePick | Value-investing screener with LLM-powered thesis generation | 🔨 Building |
Idea → Claude Code / Lovable → Next.js + TypeScript → Vercel
Multi-LLM by design: Claude for the hard reasoning, OpenRouter to route everything else to cheaper models.
The gap between "we have an AI strategy" and "our AI product has 10,000 users" is where I work.
Real users. Real revenue. AI that earns its place in the business.
📍 Paris, France