BETA · privacy LLMs & voice servers operational · GPU upgrade underway for faster responses · packages may still change Status & Roadmap →

Knowledge storage & documents: how the AI learns from your business information

Upload your manuals, FAQs and product documentation and the AI learns from them instantly. Discover how the knowledge base works, which file formats are supported and how to maximise answer quality.

← ← Back to knowledge base

How does the AI learn from your documents?

The Einstein AI platform uses Retrieval-Augmented Generation (RAG). You upload your own documents — manuals, FAQs, product sheets, policy documents — and the system processes them into a searchable vector database. When a user asks a question, the system finds the most relevant passages from your documents in milliseconds and sends them as context to the language model, which then generates an answer based on your specific knowledge.

Supported file formats

How to optimise your knowledge base

  1. Use clear headings and sections — well-structured documents yield better search results.
  2. Add an FAQ document with the 20–30 most common questions and concrete answers.
  3. Keep documentation current — outdated information leads to incorrect answers.
  4. Use source labels so the bot can cite its sources, building user trust.
  5. Split large documents into focused chapters for better retrieval precision.

Privacy and security of your documents

Your documents are stored in an isolated per-tenant environment. No other platform customer can ever access your knowledge base. You can delete documents at any time via the dashboard.