I build AI agents.your automations.your AI product.platforms.your MVP in 72h.AI agents.

An AI Product Builder designs, automates and ships a digital product end to end by orchestrating AI agents. They cover product framing, design, specs and build, where a classic team chains a PM, a designer and developers. They don't ask a chatbot questions. They run agent workflows that they brief, verify and correct. I do this job on my own platforms: sceniq.studio, tondroit.fr and DevisGo.

An AI consultant identifies the tasks where artificial intelligence delivers a measurable gain, frames the use cases, picks the tools and drives the rollout to production. Most deliver a report and a roadmap. I deliver the system, because I build it myself with Claude agents. On the Paris market a senior AI consultant charges €600 to €1,200 a day. I work on a fixed fee, with something running at the end.

Repetitive tasks built on text or data. Interview and research synthesis, writing specs and user stories, generating and testing code, producing content and video, sourced answers to customer questions, generating quotes and documents. According to Anthropic's State of AI Agents Report 2026, 57% of organisations already deploy agents on multi-step workflows. Framing the problem, making the calls and owning the result do not get automated.

Rarely because of the model. The causes are organisational. Vague scope, unprepared data, underestimated integrations, no human checkpoint. A poorly framed agent fails fast and with confidence. I work in the opposite order. Framing and success criteria first, a working prototype tested on your data in 72h next, the deployment sprint only after that.

A Claude expert designs working systems around Anthropic's models. Agents with the Claude API and the Agent SDK, MCP connectors, reusable skills, sourced RAG, Claude Code to ship production code. On an engagement, I take a business process and turn it into an agent workflow with its verification points. I run more than 100 Claude skills and 16 agents in production on my platforms.

An AI agent orchestrator breaks a process into tasks, hands each task to a specialised agent, connects the agents and puts a human checkpoint where a mistake is expensive. They write the briefs, define the success criteria, verify the outputs and correct them. This is the role I hold on my platforms and on client work. The rare skill is not starting an agent. It is knowing when to stop it. More on AI agent orchestration →

Three cases come back. A company wants to automate specific tasks with Claude agents and doesn't know where to start. A product team wants a working AI prototype before committing a development budget. A founder wants a complete AI product without hiring four people. I work freelance, in Paris or remotely, on a fixed fee, with a deliverable that runs at the end of every engagement.

From €3,500 up to €6,000 ex. VAT depending on complexity and the data to connect. The sprint covers framing the hypothesis, a working prototype, agent or product, a test with real users or real data, and a costed go or no-go recommendation. The fee is fixed before we start, never time and materials.

A classic agency puts a salesperson, a project manager, a designer and two developers between you and the result. Every relay costs time and information. Here one brain frames, designs, has the agents write the code and ships. The fee is two to three times lower for the same scope, and you talk to the person doing the work. The limit is real too: I take no more than two engagements at a time.