The knowledge base that keeps your agent from making things up
Upload FAQs, answers, and policies: your on-site AI agent retrieves from this base using RAG instead of guessing an answer.
What you feed in is exactly what it answers with
Not crawling, not guessing — retrieval from an approved source
- Every answer the agent gives is retrieved from what you've uploaded to the knowledge base — never from free-form guessing or the open internet
- Shipping, return, and warranty policies are documented once and automatically surface in every relevant customer answer
- Editing content in the knowledge base affects the agent's next answers immediately, with no separate approval step required
- Reviewing the agent's past answers reveals content gaps, and closing those gaps directly improves the accuracy of future answers
Knowledge base capabilities
A smart knowledge foundation for your on-site AI agent
How do you build a knowledge base?
3 steps to a precise knowledge base
Add your most common FAQs, shipping and return policies, and basic product information — this is the foundation the agent starts answering from.
Sort content into categories like shipping, returns, and products, so specific items are easy to find and update later.
Review conversations the agent has handled, find questions it didn't answer accurately, and add new content to the base so accuracy keeps improving over time.
Knowledge base management screen
The knowledge base view lists questions and answers organized by category, with quick inline editing and a flag showing which entries the agent relies on most.
Frequently asked questions
There's no limit on content volume — the agent can retrieve from as large a knowledge base as you build. The more comprehensive the base, the more varied the questions the agent can answer accurately.
Build an accurate knowledge base for your agent
Start from the ready-made templates and refine accuracy over time