Paul Jeandel

Most AI projects stop at the demo that impresses in meetings, then break as soon as they are connected to the real world. My job starts where things get serious: making sure Claude works for real, on your data, in your processes, reliably.

I am a data/AI engineer. This changes how I work: I treat an AI need like an engineering problem — we scope, choose the right model for each task, measure, and make it maintainable. No black boxes, no "magic," just systems we understand and keep under control.

What sets me apart: I don't just use Claude, I build with it. I've developed my own plugin and I live daily in Claude Code, MCP, agents, and sub-agents. In short, I can connect Claude directly to your tools and your data — not just deliver a slightly clever chatbot.

Where I can help you:

Design end-to-end AI system architecture, from scoping to deployment, with the right human checkpoints.

Develop custom solutions: agents, sub-agents, MCP servers, Claude Code plugins, internal tools connected to your data.

Automate what costs you time: n8n / Make workflows, API integrations, orchestration.

Get your team using Claude / Claude Code for real: onboarding, structured first uses, advanced prompt engineering.

Audit existing systems: identify what's costly, what's drifting in quality, and where you're too dependent on a vendor.

I work with multiple models when the project requires it (Claude first, but also GPT, Mistral, or Gemini depending on cost, performance, or compliance constraints). The goal is never to shoehorn AI everywhere: it's to solve your problem, sometimes with less technology than expected.

Talk to me about where you are — I'll tell you frankly what's feasible, useful, and what's not worth the effort.