How it works
RAGForge is being built to connect the questions people ask with the content you already maintain.
See the experience first ↗Choose the knowledge
Choose the website pages or documents your assistant should use. Begin with one topic and the content you trust.
“Help new customers set up our product.”
Specific enough to test. Useful enough to matter.Find, then answer
Retrieval finds relevant material in your sources. The assistant uses that material to compose an answer and point back to the evidence.
That’s retrieval-augmented generation, or RAG. The practical benefit: answers connected to your content.
Review before sharing
Check whether answers are useful, whether sources support them, and what happens when the content is missing. Use those findings to decide what is ready to share.
A convincing answer still needs evidence.
Evaluate real questions from your audience.Before you start
This website offers an interactive example and a local assistant brief builder. Self-service uploads, live chat, payments, and deployment are not available through this preview.
The intended assistant has a defined collection of sources and a specific job. It retrieves relevant material from that collection to support its answers, rather than relying only on general model knowledge.
Private knowledge requires a deployment with the right access controls, retention rules, and audience restrictions. Those need to be agreed before connecting private content. This preview does not accept document uploads.
No. Your documentation remains the source of truth. The assistant gives people another way to find and understand it, with links back to the original material.
Define your first audience, sources, and questions.