IT solutions architect at the City of Montréal. Based in Longueuil, Québec. Close to twenty years building web applications — first document management and publishing Québec's legal texts at Irosoft, then six years in asset and wealth management at Desjardins, and since 2017 the City's application portfolio. Trained at UQAM in computer science and software engineering.
Twenty years means I've seen enough technologies come and go to be wary of hype cycles. What's happening with AI agents since 2023 is different. This isn't another framework — it's a shift in how we write software, in what counts as "working," in what can be delegated and to whom.
Why this project
Most of the writing about agents falls into one of two traps: marketing enthusiasm that oversells the magic, or disappointed criticism that ignores what actually works. Neither is useful to a developer who has to ship something Monday morning.
97 Things Every Agentic Programmer Should Know is the collection I wish I'd read two years ago. Short, concrete lessons drawn from what breaks in production. Written for the developer building agentic systems — and for the developer using one every day. No meta-discussions about the future of AI, just principles you can apply this week.
The title is a nod to O'Reilly's 97 Things Every Programmer Should Know collection. Same format: short essays, each built around a single idea, with no required reading order.
How the content is made
This site is written in collaboration with Claude (Anthropic) — which is itself a demonstration of the subject. The ideas, the editorial decisions, and the arc of each essay come from me. The drafting, the debugging, and the iteration happen in pair with the agent. That's how I work now, and that's what makes it honest to write about the subject.