Solutions · IT and platform teams

An AI platform your IT team runs itself.

Not one more SaaS you cannot control: you install RAGSuite with one CLI on your own infrastructure, connect existing sources and choose the language model. API, webhooks and system status are included, and the Community Edition is open source.

Install the Community Edition Request a call →

Linux, macOS or WSL2 · native or with Docker · no phone-home

admin@ihr-server: ~

$ npm install -g @ragsuite/ragsuite@latest
$ ragsuite init # native, or --docker
↓ checking prerequisites · Node · Python · Postgres · Redis
$ ragsuite start
✔ api :9090 · web :9191 running
→ http://localhost:9191: no data left the building.

Architecture

What you run: the RAGSuite architecture.

RAGSuite consists of a gateway on FastAPI, PostgreSQL, Redis and a vector store. The CLI brings the whole stack onto your infrastructure, natively or under Docker. You need a Linux, macOS or WSL2 host with Node 18 or newer.

You apply updates yourself, when it suits you. There is no outbound connection you do not set up yourself: no phone-home, no telemetry, offline licence in Enterprise.

System statusYour server

Application & APIFastAPI · :9090running

Web interfaceReact · :9191running

DatabasePostgreSQLrunning

CacheRedisrunning

Search indexChromaDBrunning

Language modelOllama · localrunning

0 outbound connections to the vendor

What your team gets

Operable from day one.

Install via CLI

npm install, ragsuite init, ragsuite start, on your server in minutes.

REST API and webhooks

Create and revoke keys, endpoints for search and retrieval, webhooks for events.

System status and 2FA

Live status for gateway, PostgreSQL, Redis and vector store; two-factor sign-in.

Roles and teams

Organisation, teams and roles; Google sign-in in Enterprise and the RAGSuite Engine.

Integration

Your sources, your models, your stack.

Crawl websites, upload files, connect SharePoint, Confluence, Google Drive, Notion, Slack and Gmail, more tools via MCP. As a language model you use any OpenAI-compatible endpoint, or locally Ollama and vLLM: set per project. Your keys stay with you.

Connectors →For developers →

AI Search via API

$ curl -X POST https://ragsuite.ihre-firma.de/api/v1/search \
    -H "Authorization: Bearer $RAGSUITE_API_KEY" \
    -d '{"query": "notice period framework contract"}'
✔ 200 · answer + 2 sources · 1.4 s
{ "answer": "…", "sources": [ { "title": "Logistics_framework.pdf", "page": 7 } ] }

Security and operations

Operations your security concept already knows.

RAGSuite behaves like any other application in your data centre: you back up database and index with your usual backups, and you control access through roles and two-factor sign-in. The audit log shows questions, answers and changes; in Enterprise in full and with export.

Conversation storage can be switched off, and you set deletion periods per project. That keeps data protection and the works council involved from the start.

Audit logEnterprise

All actionsLast 30 daysExport

Time	User	Action
24 Sep 09:14	a.weber	Source added
24 Sep 08:52	a.weber	Role changed
23 Sep 17:30	p.neumann	Model switched
23 Sep 11:05	a.weber	Retention changed
22 Sep 16:48	l.hoffmann	Sign-in
22 Sep 10:12	a.weber	Export created

Plans

Run the AI platform yourself or have it run?

Criterion	Community	Enterprise	RAGSuite Engine
Who runs it	your team	your team, with support	RAGSuite
Where it runs	your servers	your servers, air-gapped too	data centre in Germany
Language models	all, local too	all, local too	large models and EU providers
Roles, teams, audit log	audit log 15 days	✓ full	✓ full
Price	free	on request	from €39 a month

All pricing →RAGSuite Engine →

Start

Productive in four steps.

Install

Community Edition via CLI on a test server.

Connect

Set up the first source and language model, create API keys.

Embed

Widget into the intranet, or search via API into your own apps.

Operate

Monitor system status, apply updates, move to Enterprise when needed.

Install the Community Edition →

FAQ

Questions from IT teams.

What does RAGSuite need to run?

A Linux, macOS or WSL2 host with Node 18 or newer, plus PostgreSQL and Redis. The open source CLI brings the whole stack with FastAPI, PostgreSQL, Redis and a vector store onto your infrastructure, natively or under Docker. For local language models, also plan a suitable GPU for the model size.

Can we keep everything on our own network?

Yes. With local models via Ollama or vLLM and the offline licence in Enterprise, RAGSuite needs no internet connection. There is no phone-home and no telemetry. You apply updates yourself. That way RAGSuite also runs in isolated, air-gapped networks without any outbound connection.

Which language models can we use?

Any model behind an OpenAI-compatible endpoint, plus local models via Ollama or vLLM. You set the model per project and can switch without re-ingesting sources. Your API keys stay in your instance, and your data does not train anyone else’s model.

How do we integrate RAGSuite into our own applications?

Through the REST API with endpoints for search and retrieval, webhooks for events and MCP for other AI tools. You create and revoke API keys yourself. On top come widgets for search and chat that you embed with one snippet in your intranet or website.

What if we decide not to run it ourselves?

Then use RAGSuite Engine: we run the platform in a data centre in Germany, with the full feature set and a fixed monthly price per plan, from €39 a month. We plan a move between the RAGSuite Engine and your own server together with you, so that sources and settings are kept.

Mihaela Angelova

Delivery & Onboarding

Install RAGSuite today.

Three commands, your own server, no sign-up with us. For Enterprise, support and architecture questions, talk to Mihaela.

Install the Community Edition Request a call →