An agent is only as good as what it knows. The knowledge base is where you feed the partner support agent your websites, snippets, and documents - and connect live tools - so its answers are accurate, current, and specific to your program.
What it does
The knowledge base is where you manage everything the partner support agent draws on. You can crawl websites so the agent learns from your public content, write snippets to teach it specific facts or answers, and upload documents like PDFs as sources. The agent also uses live, portal-scoped data, announcements, tiers, commissions, assets, and CRM objects, automatically, scoped to each partner. For richer capabilities, you can connect MCP servers so the agent can call live tools beyond your content. You control every source, so you decide what the agent knows and can keep it current as your program changes.The problem it solves
A generic agent gives generic, wrong answers:- Answers are vague or made up → Grounded sources keep answers accurate.
- Content is scattered → Websites, snippets, and documents in one knowledge base.
- Knowledge goes stale → Re-crawl and update sources as things change.
- The agent cannot reach live systems → MCP servers give it live tools.
From problem to solution
The knowledge base puts you in control of the agent’s brain. Add the right sources, keep them current, and connect live tools, so the partner support agent answers like an expert on your program.Use cases
Ground the agent
Feed it accurate sources.
Reuse existing content
Crawl sites and upload docs.
Teach specific answers
Add snippets for key facts.
Extend with tools
Connect MCP servers.
Impact
- AI-native - The agent answers from sources you control.
- Low total cost of ownership - Partner ops manages knowledge with no engineering.
Who it’s for
Partner Operations
Curates and updates knowledge sources.
Partner Marketing
Adds content and key answers.
Works with
Partner Support Agent
Grounds the partner support agent’s answers.
Asset Library
Indexes your library so answers cite real content.
MCP
Exposes the same knowledge to AI clients.
Headless
Going deeper
How to
Setup, configuration, and all how-to guides.
API reference
Integration surface and code.