Ragen
What it is
Ragen is a self-hosted knowledge assistant that answers questions from company documents and tools with citations to their sources. Organizations can create separate assistants for departments such as HR, sales, and compliance. Its website chatbot handles product and policy questions and can be embedded on allowed domains with one line of code. Ragen exposes company knowledge through MCP, an OpenAI-compatible API, and a TypeScript SDK. Listed integrations include Google Drive, HubSpot, ClickUp, Fireflies.ai, Google Calendar, Gmail, Slack, Notion, Google Analytics, WooCommerce, Open Mercato, and Rejestr.io; BaseLinker and KSeF are marked as coming soon. Ragen Brain organizes scattered documents into knowledge pages, flags contradictions and outdated files, and requires approval before publishing. Permissions are checked per file and folder for each answer. Documents, vector storage, and conversation history remain on the customer’s servers or cloud account. The free Self-hosted plan has no license fees, but customers pay for infrastructure and models.
For and against
For
- Answers include source citations
- Separate assistants support departments
- Self-hosted data placement
- Per-file and folder permission checks
- Free plan has no license fees
Against
- Customers pay for infrastructure and models
- Two listed integrations are coming soon
- Focused on document-based question answering
The verdict
Ragen combines cited document answers, department assistants, integrations, and self-hosted control. Its free plan removes license fees, while infrastructure, model costs, and the product’s document-focused scope remain important considerations.
Who it suits
Ragen suits organizations needing an on-premise or self-hosted assistant for HR, sales, service, operations, or partner teams. It is intended for controlled document-based answers rather than image generation, voice, or code execution.
Plans and prices
Self-hosted
Docker image on your infrastructure · RAG assistant · Department assistants · Website chatbot
Read 3 Oct 2026 · ragen.ai
Enterprise
Self-hosted features · Hands-on rollout · Custom MCP integrations · White-label chatbots
Read 3 Oct 2026 · ragen.ai
On the record
- Access control
- Permissions are checked per file and folder for each answer, including answers requested through AI agents.ragen.ai · read 3 Oct 2026
- AI agents
- Ragen exposes company knowledge through MCP, an OpenAI-compatible API and a TypeScript SDK.ragen.ai · read 3 Oct 2026
- Data hosting
- Ragen runs on the customer's own servers or cloud account, and documents, vector store and conversation history stay within that infrastructure.ragen.ai · read 3 Oct 2026
- Department assistants
- It supports separate assistants for teams such as HR, sales and compliance.ragen.ai · read 3 Oct 2026
- Integrations
- The integrations page lists Google Drive, HubSpot, ClickUp, Fireflies.ai, Google Calendar, Gmail, Slack, Notion, Google Analytics, WooCommerce, Open Mercato and Rejestr.io.ragen.ai · read 3 Oct 2026
- Intended users
- The site presents Ragen for company teams including HR, sales, customer service, operations and partners, and says it is aimed at organizations needing an on-premise or self-hosted knowledge assistant.ragen.ai · read 3 Oct 2026
- Knowledge preparation
- Ragen Brain turns scattered documents into knowledge pages, flags contradictions and outdated files, and requires human approval before publishing.ragen.ai · read 3 Oct 2026
- Limitations
- The site says Ragen is for answering questions from documents with access control, and recommends general-purpose tools for image generation, voice or code execution.ragen.ai · read 3 Oct 2026
- Model choice
- Customers can use models locally, through their own provider keys or through an OpenAI-compatible endpoint.ragen.ai · read 3 Oct 2026
- Not yet available
- The integrations page labels BaseLinker and KSeF as coming soon.ragen.ai · read 3 Oct 2026
- Purpose
- Ragen answers questions from a company's documents and tools, with citations to the source.ragen.ai · read 3 Oct 2026
- Security
- Conversation content can be encrypted with keys in the customer's AWS KMS or Scaleway Key Manager, and optional masking covers PESEL, NIP, REGON and card numbers.ragen.ai · read 3 Oct 2026
- Support
- Self-hosted includes deployment documentation and community support; Enterprise includes an SLA, priority support and a dedicated Customer Success Manager.ragen.ai · read 3 Oct 2026
- Usage controls
- The admin panel supports hard monthly organisation limits for tokens, cost or messages, plus team-level request-rate and model limits.ragen.ai · read 3 Oct 2026
- Website chatbot
- The website chatbot answers product and policy questions 24/7 and can be embedded with one line of code on allowed domains.ragen.ai · read 3 Oct 2026
