Ask instead of search.
One search that finds everything — and shows only what you're allowed to see.
Type what you need — in plain language. The search looks everywhere at once: documents, notes, emails, files. You only get what you have access to. No more hunting through ten programs.
From everyday work
"Which orders are open this week?" — and the answer comes right away
Instead of opening three programs and filtering, you just ask in chat. The answer comes from your own data, with sources, in one sentence. The assistant is trained on your own world — it knows your terms and workflows, not just generic phrases. Anyone who wants more clicks into the details — anyone who just needs the number already has it.
How it fits together technically +
A federated search indexes every source and filters per document by permission — nobody sees what they shouldn't. On top sits an assistant that answers in natural language (RAG over your own docs) and can read documents. Every building block is also available as an MCP server for AI agents. The language models run on-prem behind a proxy — data stays in-house, GDPR-safe. Recurring flows — notifications, hand-offs, escalations — are orchestrated by a visual automation (n8n) that wires in the same building blocks as tools: "ask" turns into "done automatically".
Technology used
What it's built on
A lightning-fast search that understands what you mean — even with typos. Scans documents, notes and files all at once.
The database everything sits on — the standard for 30+ years. It even stores the "meaning" of text so the AI finds the right answers.
Also used by: Apple, Spotify and Instagram build on it.
Connects your programs and handles recurring tasks on its own — no coding.
Also used by: Open source from Berlin, used by thousands of teams.
The same class of language AI as the well-known chatbots — but it runs in your house, not someone else's cloud.
Also used by: European AI from Paris.
Live demo
Open demo ↗