Question 1

How much does an AI integration project cost?

Accepted Answer

The budget depends on scope: a targeted use case (an automation, a simple RAG assistant) stays accessible, while agent orchestration or dedicated infrastructure represents a larger investment. We always provide an estimate after a diagnosis of your needs, favoring use cases with strong return on investment.

Question 2

What types of AI services do you offer?

Accepted Answer

We work across five families: AI visibility (SEO, GEO, AEO), automation and AI workflows with n8n, the design of custom agents and RAG systems, the audit and security of your AI and web stack, and sovereign AI infrastructure. Each service is based on what we operate ourselves in production.

Question 3

What are your timelines for an AI project?

Accepted Answer

Timelines depend on the scope. A targeted automation or a first RAG assistant can go into service within a few weeks, while agent orchestration or a full infrastructure takes several months. We work in short iterations to deliver value early and measure it.

Question 4

How do you measure the ROI of an AI project?

Accepted Answer

We define concrete indicators with you upfront: time saved, cost per task, volume processed, visibility gains. We rely on benchmarks from our own operations, such as an editorial pipeline at around €0.46 per article or an image pipeline cutting costs by about 72%, then track results over time.

Question 5

How do you handle GDPR and the AI Act?

Accepted Answer

We frame every project from a compliance angle: data minimization, traceability, and hosting on a sovereign infrastructure so your data stays under your control. We account for AI Act obligations according to the risk level of your use case. Final legal validation remains your responsibility.

Question 6

What is the difference between an AI agent and a simple chatbot?

Accepted Answer

A chatbot answers questions from a model. An agent goes further: it chains actions, uses tools, queries your data through a RAG engine and can complete a task end to end. We operate an orchestration of one orchestrator and fourteen specialized agents, with guardrails and human validation.

Question 7

What is RAG and why do you need it?

Accepted Answer

RAG (retrieval-augmented generation) lets an assistant answer from your own documents rather than its general knowledge alone. This reduces errors and makes answers sourceable. We operate a hybrid RAG engine with 51,912 indexed fragments.

Question 8

How do you ensure security and sovereignty?

Accepted Answer

We host models, data and services on our own infrastructure, around 80 containers and 14 PostgreSQL databases, to limit dependence on third-party providers. We audit access, isolate sensitive services and apply controlled exposure of tools, including via MCP.

Frequently asked questions about our AI projects

Do you have other questions about your AI project?