French public services
Led product and engineering teams building data pipelines, decision tools, and grant services used across public institutions.
Build AI systems that hold up in real operations.
We are an AI product and engineering practice based in Brussels.We connect AI to the data and software your team already uses, then test it against the work it has to do.
A working prototype is only the beginning
To use an AI workflow every day, people need reliable data, clear access, links to existing software, and a plan for when it fails.
The goal is more efficient and enjoyable work
The AI becomes part of the workflow, connected to the data and software the work depends on. People spend less time copying information and checking routine results, and more time acting on what matters.
Clear permissions, approvals, and ownership let the AI act safely across the CRM, case-management tool, document store, inbox, and internal APIs. People know what it may do, who reviews changes, and who owns it.
Results are measured against the previous process, giving leaders a clear view of what works and where to improve or expand it. Teams can act faster, share ownership, and carry less mental load.
Map the route to production. Usually 2 to 4 weeks.
We start with one workflow, prototype, or existing AI system. We follow the work as it happens, find the dependencies and decision points, then show what must change before people can rely on it.
You leave with
- maps of the current work, target workflow, systems, data, permissions, and owners
- a baseline, a set of real cases, known risks, and release criteria
- a costed recommendation to build, buy, narrow, defer, or stop
Sometimes the right answer is not to build. Better to know that before paying for another prototype.
After a short call, we quote a fixed fee for one workflow. We need access to the people who do the work and a useful sample of real cases. Model, software, and other third-party costs are separate.
From first question to live system
- 01
Frame the operation.
- 02
Map how it works today.
- 03
Decide what AI should and should not do.
- 04
Build the smallest complete system.
- 05
Verify it with real cases and users.
- 06
Operate it with clear ownership and responsibilities.
Experience with systems people depend on
Hugues Maignol brings 15 years of product and engineering experience across public services, insurance, telecommunications, and large digital platforms.
Insurance claims
Designed a claims-intake and evaluation system that prepared smaller cases automatically and routed complex ones to experts.
Telecommunications
Built real-time data systems and sales dashboards used on the web, mobile devices, and shared displays.
Could this improve how your organisation works?
Tell us what happens today, which systems are involved, and where the work gets stuck. No polished brief needed. We'll tell you plainly whether we can help.