RAG engine: the knowledge engine for your data.
A RAG engine connects your documents, websites and applications, finds the matching passages for every question and lets a language model word an answer from them, with a source on every claim. RAGSuite Engine is that knowledge engine: hosted in Germany or on your own server.
Community Edition free · RAGSuite Engine from €39 a month
Key takeaways
- A RAG engine answers from your own data, not from general model knowledge, and names the source for every claim.
- It has three parts: connect data, retrieve the matching passages, answer with sources.
- Chat, search, widget and app are output routes of the engine, as are the MCP server and the REST API for AI tools.
- RAGSuite Engine runs in Germany, from €39 a month. You run the Community Edition yourself for free.
What is a RAG engine?
RAG stands for retrieval-augmented generation. A RAG engine is the software that does it: it finds the matching passages in your data for a question and hands them to a language model. The model answers on that basis and names the sources. We call the result a knowledge engine.
The idea behind it: a language model does not know your company. If you give it the right passages, it answers concretely and checkably instead of guessing. The engine is the brain for your data. People and AI tools query it.
Connect, retrieve, answer: how the engine works.
Connect.
Upload documents, crawl websites, connect applications through connectors. The engine splits the content into passages and makes it searchable. Changes in the sources are picked up.
Retrieve.
For every question the engine searches by meaning, not only by keywords, and picks the best passages. This is the part that decides the quality of the answer.
Answer.
A language model words the answer from those passages. Every claim points to the document it comes from. If the engine finds nothing suitable, it says so.
Knowledge engine, enterprise search, chatbot and language model compared.
| What it does | Answer with a source | Knows your data | |
|---|---|---|---|
| Knowledge engine (RAG engine) | finds passages and words an answer from them | ✓ on every claim | ✓ from your sources |
| Enterprise search | returns a list of hits | hits, no answer | ✓ from your sources |
| Chatbot | holds a conversation | only if connected to an engine | only if connected |
| Language model | generates text from training knowledge | no | no |
A chatbot is an interface, a language model a building block. The engine is what supplies both with your knowledge.
One engine, many routes to the answer.
Chat
The AI chatbot answers questions from your data, with a source in every answer.
AI chatbot →Search and widget
AI search summarises hits. As a widget it runs on a website or intranet.
AI search →MCP server
AI tools that speak MCP query the engine and receive passages with their sources.
MCP server →REST API
Your own apps and automations use search and chat through the API and webhooks.
For developers →Connectors
SharePoint, Confluence, websites and files supply the knowledge the engine needs.
See connectors →Mobile app
The app brings the same knowledge engine to the smartphone.
Mobile app →Where the engine runs: RAGSuite Engine or self-hosted.
In the RAGSuite Engine we run the knowledge engine for you, in a data centre in Germany. There are the plans Starter, Business, Premium and Enterprise, from €39 a month plus VAT with yearly billing. You can try Starter free for 7 days, no credit card needed.
If you want full control, install the Community Edition on your server. It is licensed under Apache 2.0, free, and includes the whole engine with chat, search, widgets, MCP and API. It is not either-or: teams often start hosted and move later.
For IT teams, AI teams and companies that want to put their knowledge to work.
IT and AI teams that already use AI tools need a source those tools can answer from correctly and with proof. That is what the MCP server is for. Owners and managers in small and mid-sized companies start with the hosted engine and need no server of their own.
Where it fits less: pure text generation unrelated to your data. For that you do not need a knowledge engine, only a language model.
Questions about the RAG engine.
What is a RAG engine in simple terms?
A RAG engine finds the matching passages in your own documents and applications for your question and lets a language model write an answer from them. Every claim names its source. So the AI answers from your knowledge instead of general knowledge, and can be checked.
How does a knowledge engine differ from a chatbot?
The chatbot is the interface where you ask questions. The knowledge engine supplies it with knowledge: it connects your data, finds the matching passages and passes on the source. Without an engine, a chatbot answers only from model knowledge and does not know your documents.
Is RAGSuite Engine the same as the Community Edition?
Both are the same knowledge engine. You run the Community Edition yourself, free, under Apache 2.0. We run the RAGSuite Engine for you in a data centre in Germany, with plans from €39 a month plus VAT with yearly billing. Chat, search, widgets and app come with every variant.
Can AI tools query the engine?
Yes, through the MCP server and the REST API. AI tools that speak the Model Context Protocol query the engine and receive matching passages with their sources. Your own applications use the REST API with webhooks. Both are available in the Community Edition and in the RAGSuite Engine.
Where is my data with the RAGSuite Engine?
The RAGSuite Engine runs in a data centre in Germany, operated by RAGSuite. If you want to keep the data entirely in your own environment, install the Community Edition on your server, even offline with a local model. Details on data processing and security are in the Trust Center.
See the knowledge engine with your own data.
Start a 7-day free trial or let Jürgen show you the engine on your own use case.