It is every buyer's legitimate objection to AI. Our answer is not a promise, it is an architecture: the deployment level is chosen according to the sensitivity of your data — and it can evolve with you.
Your data is never used to train AI models. It is contractual, at every level — and it is the fundamental difference with consumer tools.
Our assistants are tied to no vendor: we change engine without changing tool, for you or your data. The same requirement you put in your own supplier contracts.
Caps per company and per user, usage tracking, weekly reset: no surprise bill, no uncontrolled use.
The best models in enterprise offerings, chosen and replaceable at any time: no training on your data (contractual), art. 28 GDPR DPA, Data Privacy Framework-certified providers, short retention. Never the consumer versions. The right level for the vast majority of procurement files.
A European vendor's model, inference and data hosted in the European Union — storage jurisdiction contractually guaranteed — zero-retention option. Your documents are no longer processed outside Europe. For sensitive supplier files, strategic RFQs, public procurement.
The entire chain — application, data and model — on ANSSI-qualified SecNumCloud French cloud, the framework the French State applies to its own sensitive workloads. No non-European actor at any step, with a dedicated instance per client. For defence, nuclear, strategic pricing.
The right reflex: start at the level that matches your actual data, not at the maximum on principle. Most of our clients work perfectly well at L1 — and move a specific perimeter to L2 or L3 when a file requires it.
Consumer versions may use conversations to improve their models, and your prices circulate there with no framework or traceability. Moving to a professional offering changes the contract: zero training, short retention, controlled perimeter — for a cost far below the risk.
They are stored in your partitioned space, and sent to the AI engine for the duration of processing — with a professional provider under a zero-training contract (L1), only in Europe (L2), or on qualified sovereign cloud (L3). At every level, we can tell you exactly who processes what, and where.
Designed for: the assistants depend on no vendor, engine substitution is the priority remedy written into our contracts, and your data and conversations remain exportable. No single dependency — just like in your own panels.
Each user has their own account and conversations. Assistants are enabled company by company and user by user. You keep control of the rights — and we control none of your content: your data is not our data.
No. The level is chosen per perimeter and can evolve: L1 for the day-to-day, L2 for a sensitive file, L3 for a specific programme. The tool stays the same — it is the architecture underneath that changes.
CISO, DPO, management: bring your requirements — we show you the architecture that answers them, level by level.
