​
What Is Knowledge?

In Arahi, Knowledge is your centralized, dynamic context bank that your agents and tools can tap into for accurate, on-domain responses. It’s like giving your AI workforce access to your smartest person’s brain—structured and instantly retrievable. Knowledge can exist as:
  • Snippets: Style guides, tone examples, or response formats to shape agent behavior.
  • Databases: FAQs, product specs, procedures, or policy details for factual accuracy.
  • External sources: Content ingested from PDFs, websites, or integrated systems like Notion, CRMs, or internal docs.

​
Why It Matters

  • Grounded answers: Agents reference actual knowledge instead of relying on pre-trained model hearsay. This improves reliability and reduces hallucinations.
  • Domain expertise: Easily turn your brand voice, SOPs, and documentation into agent-ready knowledge.
  • Update once—reflect everywhere: Make a change in your knowledge base, and all connected agents and tools are instantly refreshed.

​
How to Build Knowledge in Arahi

  1. Manual input: Create structured tables with columns like Question, Answer, Category.
  2. Upload existing content: Convert CSVs, PDFs, Excel, JSON, or even audio transcripts into searchable knowledge.
  3. Web extraction: Pull in text from websites, wikis, or documentation sites.
  4. Integrations: Sync up Notion, Google Docs, CRMs, or other platforms for live context updates.

​
Using Knowledge with Agents & Tools

  • Connect your knowledge base to agents via the agent configuration panel.
  • Retrieval styles:
    • Semantic search: Finds relevant info based on meaning—not just keywords.
    • Hybrid search: Combines semantic matching with exact filters for precision.
  • Advanced retrieval patterns:
    1. Use LLM-based query refinement.
    2. Perform vector search on your knowledge.
    3. Validate and summarize results via an LLM step.

​
Quick Setup Guide

  1. Go to the Knowledge Library in Arahi.
  2. Choose how to add content: manual, upload, extract, or integrate.
  3. Structure your knowledge for readability and retrieval.
  4. Link it to your agents or tools and define how it should be queried.
  5. Test scenarios to confirm agents reference the right context.

​
It Makes your Agents Smarter

With Knowledge, your agents don’t guess—they know. You inject brand, process, and policy directly into their thinking, creating consistent, accurate, and trustworthy interactions, across any use case.