Scaling AI in aerospace and rail: takeaways from VivaTech

The Curiosity Team

• 2026-06-26

A Capgemini-hosted panel on why so much industrial AI stalls between pilot and production, and what it takes to get past it.

Last week at VivaTech, Curiosity founder and CEO Leon Zucchini joined a Capgemini-hosted panel on scaling AI for aerospace and rail. He shared the stage with Cheikh Diop (Airbus), Eric Gadzinski (Alstom), Patrick Bennani and Thierry Talucier (Capgemini), in a session put together by Romain David.

The premise: almost everyone is building with AI, few projects reach production at scale, and the ones that do tend to deliver less than the pilot promised. So the panel stayed on one question: what has to be true for industrial AI to work in production?

Four takeaways.

1. Start with the business problem. Scaling agentic AI begins with a concrete problem worth solving: a bottleneck, a recurring failure, a slow handoff between teams. The problem decides whether anything ships.

2. Graphs give agents structure. Point an agent at terabytes of flat, densely cross-referenced engineering data and it latches onto something plausible and wanders off. Industrial documents are full of technical references that embeddings handle badly. A knowledge graph gives the agent the connections between records, so it reasons over real data instead of guessing.

3. Iterate from a working use case. Spending nine months designing the ideal ontology before anything runs is the usual trap. The data is messier than the diagram (zip files, emails, fragmented sources). Sit with the business, build a narrow use case that works, and let domain experts validate it as you go.

4. Adoption and cost decide the rest. Getting engineers, technicians and quality teams to fold AI into how they work is where projects live or die. Close behind it sits the cost of running AI at scale: inference can outpace the value created, and rate limits become an operational ceiling.

This is what Curiosity is built for. It connects fragmented enterprise data into a context graph and gives AI agents grounded, permission-aware structure to work on, without replacing the systems you already run.

Thanks to Capgemini for the invitation, and to the fellow panelists for a candid discussion.