RAGSuite is a self-hosted, citation-backed enterprise AI platform: it brings AI Search, an AI Assistant and AI Connectors to your documents and business apps, running entirely on your own infrastructure. It is open core — a Community Edition under Apache 2.0 and a commercial Enterprise Edition — built for organisations that need modern AI without handing their data, or their control, to someone else’s cloud.
This is the first post on this blog, so it is the right place to say plainly what we have built and who it is for.
The problem we set out to solve
Enterprise teams want what hosted AI assistants promise: ask a question in plain language, get an answer grounded in your own documents. But for a regulated, German-speaking organisation, the usual route — upload your knowledge to a multi-tenant US cloud — is a non-starter. It introduces a new sub-processor, creates data-transfer exposure, and gives you an answer you cannot audit. RAGSuite is the other route: the same modern experience, but the data, the models and the audit trail stay under your control.
The product, in three
AI Search returns citation-backed answers across your content; AI Assistant is that in grounded chat form, with sources on every reply; AI Connectors bring your knowledge in — website crawl and upload, Gmail, MCP, n8n (Beta), and a growing Connector Marketplace, all in the open Community core. You publish Search and Assistant as embeddable widgets, so the same grounded engine serves staff and customers alike.
What makes it sovereign
It runs on your infrastructure (PostgreSQL, Redis, a vector store, optional local models via Ollama); it is model-agnostic with Compare Models; and it does not phone home — Enterprise unlocks with a signed, offline licence key.
What it costs
Pricing is deliberately simple and predictable — one annual figure your finance team can sign without a sales call.
Who it is for
RAGSuite is for organisations where “prove control” is a requirement, not a nice-to-have: regulated enterprises, the public sector, and mid-market companies in DACH and the wider EU — and the agencies who build for them, via a free-to-join partner program. RAGSuite is an innovation by NITSAN, carrying a fifteen-year enterprise track record into a product built for the sovereignty era.
Start with What is sovereign enterprise AI?, see the platform, or weigh the trade in self-hosted vs. hosted RAG. We will use this blog the way we build the product: proof over hype, specifics over slogans.
Frequently asked questions
What's the difference between Community and Enterprise?
The Community Edition (Apache 2.0, free) gives you the full pipeline — crawl, upload, the entire connector and MCP ecosystem, chat, search, widgets, all LLM providers, the REST API, basic audit and unlimited users. The paid Enterprise Edition adds SSO/RBAC, full audit and compliance exports, Compare Models, deep query tracing, advanced analytics and the mobile app (Beta). Enterprise services — white-label, air-gapped support, SLA and a dedicated CSM — are available by agreement for large or regulated deployments.
Can I run it air-gapped?
Yes. Enterprise capabilities unlock with a signed, offline licence key — no phone-home — and local models run via Ollama, so an air-gapped instance works as a first-class scenario.
Which models does it support?
OpenAI, Anthropic, Mistral, Gemini, or local models via Ollama. It is model-agnostic, and Compare Models lets you route per project.
Sources & further reading
- Apache License 2.0 — the Community Edition licence
- RAGSuite — pricing — Community / Enterprise Edition
- RAGSuite — platform — AI Search, AI Assistant & AI Connectors