I build products with AI: agents, automations, apps and SaaS.
I came to code from e-commerce. Since 2023 I've run online stores, and today most of their operations run on agents I built: customer support, ads, SEO, monitoring. Thirteen of them report to a control room I wrote myself. Most of what I know comes from watching them break.
Open to projects and consulting. If a process eats hours of your week, or an AI prototype needs to hold up in production, that is the work I take on. Remote, in English or French.
- Agents and automations that run unattended: n8n, Node.js, Python, LLM APIs, with logs and alerts so you know the day they stop.
- Internal tools and small SaaS, from the first script to a deployed product.
- E-commerce plumbing: Shopify, ad platforms, tracking, structured data.
- AdAnvil: an MCP server that lets an AI agent edit vertical video ads from your own footage. On PyPI and the official MCP registry.
- n8n-workflows: n8n templates for an online store: AI email support, site chat, comment moderation, shipping and stock alerts.
- skills: agent skills that measure instead of guess.
- StoreCanary: a SaaS that watches e-commerce storefronts for what quietly stops them selling.
- Measure before you opinionate. An agent with a ruler beats an agent with a confident essay.
- Boring and reliable over clever. If it needs babysitting, it is not done.
- Ship small, then iterate on real usage.