Applied AI & AI-augmented engineering
AI has significantly changed how I work without changing what I expect from a software system: clarity, reliability, security, and control.
I use it daily to accelerate research, analysis, product design, and software delivery. It has also brought me closer to code and deployment again—not by writing every line myself, but by orchestrating a workflow where specification, agents, review, tests, deployment, and monitoring remain under control.
The value therefore does not come from the model alone. It comes from the system built around it: context and instruction quality, problem decomposition, tool selection, guardrails, review, security, observability, cost control, and the ability to take over when needed.
- Daily use of ChatGPT, Claude, Gemini, Codex, and Lovable depending on the task and desired level of autonomy
- AI-assisted software engineering: specification, generation, review, testing, debugging, and iteration
- Orchestrating the path from idea to production: Git, branching, CI/CD, deployment, instrumentation, and monitoring
- Guardrails that preserve quality, security, maintainability, and cost control
Two concrete environments keep this approach grounded in reality: xavier-barry.fr, which I use as a lab for AI-assisted software production and deployment, and Mama Flow, a B2C product I cofounded and operate in production.