An AI chat that knows your documents — and stays in-house.
Hippo Chat is a private AI assistant for your entire organisation. It answers from your own knowledge base, runs on-premises or as a managed service in Germany — and not a single line of your data ever leaves your environment.
Your teams already use AI — just not on your terms.
When there's no safe tool available, people reach for whatever is at hand: the public chatbot in the browser. Engineering drawings, quotes, contract drafts and internal know-how end up on someone else's servers — outside your control, often outside the EU, and no longer covered by the GDPR.
And even then, generic chatbots give no usable answers: they don't know your products, your standards or your internal processes. They guess instead of knowing, and nobody can trace where an answer came from.
Hippo Chat turns that around. The assistant runs inside your environment, draws on your own documents, and backs its answers with sources. Your data stays in-house — because it never leaves.
Your own company chat in four steps.
From your documents to a rollout across the whole team — guided, not left to fend for itself.
Connect your knowledge sources
We connect Hippo Chat to your documents and systems — manuals, quotes, technical specifications, wikis. The assistant indexes them so it can later answer from your own knowledge rather than the open internet.
Choose on-prem or managed, pick models
You decide where the chat runs: fully on-premises on your hardware, or as a managed service in Germany. Together we pick the right models for your use cases and your budget.
Roll out with SSO and roles
We connect Hippo Chat to your single sign-on and set up roles and permissions per team. Each department sees exactly the knowledge sources it's cleared for — no more, no less.
Monitor, refine, expand
After launch we tune retrieval quality against real usage, add more sources and models, and support your team so it can run and grow the assistant on its own.
An open, auditable stack — layer by layer.
Every component is open source, self-hostable and auditable. No vendor lock-in, no black boxes.
Open WebUI
The familiar chat interface for your team — with histories, shared prompts and roles. Runs entirely inside your environment, with no cloud connection.
Lucene
Apache Lucene — the proven search engine that pulls answers from your own documents (full-text and vector retrieval, with source citations).
vLLM
High-throughput serving for open-weight models with PagedAttention. Handles your teams' concurrent requests efficiently on your own hardware.
LiteLLM
One unified, OpenAI-compatible way into every model. A single endpoint, full traceability and cost control across the whole deployment.
An assistant your legal team signs off on.
Concrete benefits your team feels from the first week.
"Our engineering team stopped pasting drawings into public tools. Now they ask the internal chat — it knows our standards, names the source, and everything stays in-house."
Frequently asked questions about Hippo Chat.
Straight answers on privacy, training data and access.
Ready for sovereign AI?
Let's talk about your challenges, no strings attached. Together we'll find the best solution for your needs — fast, straightforward, and tailored to you.