1. Your documents stay with you
The documents, what DeepView draws from them, the conversations and the accounts: everything is stored on your company's machine. None of it is copied over to us.
Security & privacy
Our four principles
The documents, what DeepView draws from them, the conversations and the accounts: everything is stored on your company's machine. None of it is copied over to us.
Nobody reaches anything without having identified themselves, and every action is authorised according to that person's role on the project concerned.
Not us, not a third party. Your documents serve only to answer your own questions, at the moment they are asked.
The only thing that can go out is what is submitted to the AI model, and you are the one who chooses the architecture, below.
Where your data goes
The table below deliberately contains no grey areas.
| Data | Where it is kept | Who can reach it |
|---|---|---|
| Your documents | In your environment | Authorised users of the project concerned |
| The information extracted from your documents | In your environment | DeepView only, for analysis and information retrieval |
| The knowledge base | In your environment | DeepView only |
| User conversations | In your environment | Only the users concerned |
| User accounts and credentials | In your environment | Nobody: passwords are never readable, including by our teams |
| The information sent to the AI engine | In your environment, or with the provider you have chosen | According to the architecture you have selected |
The last row is the only one that calls for a decision on your part: that is the subject of the next chapter.
The sensitive point, addressed head-on
To write an answer, an extract of your documents has to be submitted to an AI model. That is the only moment when anything can go out. We do not impose it: you choose the architecture.
No model is imposed by the product: DeepView addresses whichever one you designate, which is what lets it run on a site entirely cut off from the outside if you choose architecture 1.
The DeepView server and an open-source LLM both run on your own infrastructure. Nothing ever leaves your premises, at any point.
The server stays on your infrastructure and connects to the LLM provider or providers you have already contracted with. DeepView does not need a powerful model to work well.
The server stays on your infrastructure; the call to the model goes to a secure environment hosted in Europe and provided by Alister AI. The communication is encrypted.
Who sees what
Users are created and removed by whoever administers the system on your side. Passwords are never stored in clear text: even we could not read them.
Documents belong to a project. A person sees only the projects they are part of: the others do not exist for them.
What you ask belongs to you alone. A colleague on the same project has no access to it, and neither does the administrator.
Transparency
A vendor claiming to receive nothing at all is rarely being transparent. DeepView is designed to keep exchanges with our services to the strict minimum. We state clearly which information may reach us and which always stays in your environment.
In short: what we know is limited to the checks required for the application to work properly, and we know nothing about your infrastructure, your documents, your knowledge or your exchanges.
Compliance
Because the software runs on your machines and we do not access its content, any personal data present in your documents remains entirely under your responsibility and your control.
In concrete terms: we host nothing, we do not read your documents, and nothing is transferred to us. The precise terms (including for troubleshooting that requires temporary access) are set out in the contract.
Security questions
Write to us: security topics are handled directly by the team that builds the product, with no sales intermediary.