Our services
-
Tailor-made AI system development
Every business has its own data, constraints and specific problems, which AI can sometimes solve. We design systems suited to your context, for example:
- Generative AI (specialised assistants and agents)
- Digital twin of an industrial process
- Reduced-order models to cut numerical simulation costs
- Industrial process optimisation
We support you from prototype to production.
-
AI infrastructure
We favour AI that is more local, more frugal and sovereign: your models run on your premises, on specialised, secure infrastructure sized to actual needs. You keep control of your data, predictable costs and greater technological independence. Already running your models in the cloud? We also help you optimise their performance and costs.
- Sizing your infrastructure to actual needs, on premises or in the cloud
- Quantified assessment: performance, power consumption and total cost
- Choosing open-source models suited to your use case, and adapting them to your business
- Inference optimisation: quantisation, inference server, latency, load management
- Hybrid architecture: local for sensitive data, cloud for the rest
- Deploying your deep learning models, LLMs and AI agents
-
Available from 2027
AI system cybersecurity
AI systems open up new attack surfaces that conventional security approaches do not cover. We audit and test your systems against these specific threats:
- Jailbreaks and model misuse
- Prompt injection
- Access control
- Security of RAG, agent and Model Context Protocol (MCP) integrations
- Securing infrastructure and models
Our analyses draw in particular on recognised frameworks (MITRE ATLAS, OWASP Top 10 for LLM Applications).