Your own AI assistant that knows your documents and stays in the EU

Your team asks the same kinds of questions every day. What is the notice period in this contract? How does the returns procedure work again? Which version of the safety data sheet is the latest? The answer is somewhere: in a folder on the server, in an old email, or in the head of the one colleague who has worked there for 15 years. Meanwhile, half your people paste text into ChatGPT because it is fast, client names included.

What a private AI assistant is and is not

A private AI assistant is a chat window your team uses the way they use ChatGPT. Type a question, get an answer, have a text rewritten or summarized. The difference is under the hood. The language model runs in our managed environment in the EU, shielded for your organization; the settings, logs, and data are yours. What you type stays there. Nobody trains a model with it, and nothing goes to a provider outside Europe.

It is not a system that makes decisions on its own, and it does not replace expertise. Think of it as a colleague who has read every document and never gets tired of the same question. People still check what comes out and decide what happens with it.

Your documents as knowledge

The assistant only becomes truly useful once it knows your company. We connect it to the documents that matter: the handbook, the contracts, the work instructions, the product specifications, and the meeting minutes of the past years. When someone asks a question, the assistant searches those sources first and gives the answer with a reference to the document it came from. That way, you can check it. The technology behind this is the same as in our AI document search service. The assistant adds conversation and text editing on top.

We respect permissions. If someone may not open a folder, they get no answers from it through the assistant either. A personnel file stays with HR, and a quote in progress stays with sales. Documents you add later, the assistant knows the next day. If your documents still sit in an archive as scans, we first make them readable with Document processing (OCR + AI).

Why not just use ChatGPT

For a recipe or a vacation email, ChatGPT is fine. For a quote with client data, a personnel file, or a contract with a confidentiality clause, it is not. You do not know where the text goes, who keeps it, or whether it ends up in a later model. Under the GDPR, you are the responsible party, not the provider of the chat window. And a ban rarely helps: people use it anyway, just on their own phones.

A private assistant solves that without taking anything away from your people. They keep a chat window that works the same way, only it runs shielded for your organization. You keep the logs yourself, you decide which sources and topics are allowed, and you have a clear answer ready when a client or an auditor asks. If you also want this properly arranged on paper, with a register and a risk assessment under the EU AI Act, AI governance and audits builds directly on it.

On your team's laptops in 10 business days

We start small. You pick one department and one pile of documents. Within 10 business days, there is a prototype, and those people ask it real questions. That is the moment you see whether it is right: does it give the correct article from the handbook, or does it talk past it? What does not work, we adjust or leave out. If it works, we add more sources and open the assistant to the rest of the company, in the browser or in the tools they already use.

You own the environment, the settings, and the data. There is no per-user license that gets more expensive every year. If you later want the assistant to write up meetings as well, we add that through Transcription and summaries. And if the intake shows that a good search function is already enough for what your team needs, we tell you right away.