Case study · Portfolio
An AI-powered internal tool that compares mockups to quotes, checks RLP compliance via RAG, detects nearby Monuments Historiques, auto-fills CERFA forms, and generates cadastral plans.
The project
The context, the challenge, what we did and the results, step by step.
This case study is based on the project’s own information; any outcomes mentioned are not independently measured.
An AI-powered platform that automates the preparation of CERFA permit applications for Kappeler Publicité’s outdoor signage projects. It pairs a retrieval-augmented knowledge base of French signage regulations with Claude AI to pre-fill, check and generate the administrative paperwork each installation requires.
Installing outdoor signage in France means filing CERFA permit forms whose fields, supporting documents and local rules vary by municipality. Preparing them by hand was slow and error-prone, and a single inconsistent field could delay a whole installation.
Kappeler Publicité assembles CERFA dossiers in a fraction of the previous time, with far fewer corrections. Regulatory know-how that used to live in a few people’s heads is now a searchable system, making permitting faster and more consistent across projects.
Technology
The stack and the key features behind the project.
A Next.js/TypeScript app backed by a pgvector knowledge base of the relevant CERFA forms and regulations. A RAG pipeline surfaces the right rules per case, and Claude AI drafts and validates the fields, flagging inconsistencies before submission — the team exports a ready-to-file dossier instead of starting from a blank form.
More projects
Projects in the same category first, then other work.
Your context
Share the starting point, constraints and decision ahead. We will frame a useful first step.
Direct conversation · No imposed scope · Risks surfaced early
