Ragen AI
Articles about AI, automation and the future of work
Featured AI in the organization
RAG (Retrieval-Augmented Generation) lets AI models answer questions based on your company's documents - not generic knowledge from the internet. See how it works and where it delivers real value.
Product
Answers that read more of the document, sources that open at the passage, connectors and guardrails you set from the panel instead of a release, and a version number that means something again. A review of two weeks of Ragen AI.
Product
LiteLLM stops being a required service, the queues move to BullMQ, and the API starts speaking the OpenAI format. A review of what changed in Ragen AI, with a warning for anyone already running a stack.
Integrations
You have workflows in n8n and a knowledge base in Ragen. One step is missing: "ask our documentation". Now that assistant_id is optional, Ragen answers n8n's own OpenAI nodes. Three routes: an OpenAI credential with your own base URL, HTTP Request, and an MCP server for agents.
Product
Citations with page numbers, a ⌘K command palette, design system v2, an MCP server and the create-ragen-app installer — a round-up of changes in Ragen AI.
Governance
68% of employees use AI without IT's approval. 38% paste sensitive data. Blocking doesn't work. How to build an AI policy your people actually follow instead.
AI Strategy
MCP is the open standard connecting AI to your tools. Instead of seven chatbots, one assistant that reaches HubSpot, Gmail, Slack and Drive in a single question.
