Overview
Give your AI the two capabilities that make it useful — Endpoints (Tools) to act on your APIs, and a Knowledge Base (RAG) to ground answers in your content.
Out of the box, EgoX can hold a conversation. To make it useful for your product, you give it two capabilities — both configured per project in the Console:
- Endpoints (Tools) — external REST/GraphQL APIs the AI may call to do things and fetch live data (check an order, create a ticket, look up a customer).
- Knowledge Base (RAG) — documents the AI can retrieve from to ground its answers in your content (policies, product docs, FAQs).
Recommended: build these with an agent
The fastest, most reliable way to create endpoints and knowledge is to connect the EgoX MCP and ask your AI agent (Cursor, Claude, …) to do it. It reads your API, GraphQL schema, or docs and writes correct definitions in one shot — then you review the drafts in the Console. We think agent-built configuration beats hand-entry — fewer schema mistakes, and it scales to dozens of tools without the clicking.
Which one do I need?
| You want the AI to… | Use | Example |
|---|---|---|
| Act or fetch live, per-user data | Endpoints (Tools) | "Where's my order?" → calls getOrderStatus |
| Answer from your content (static, shared) | Knowledge Base (RAG) | "What's the refund window?" → retrieves your policy |
Both can be on at once — a single /ask can retrieve knowledge and call a tool. Which
path runs is decided by the request intent:
vanilla, rag, tools, or rag_tools.
In this section
Endpoints (Tools)
Define REST/GraphQL APIs the AI can call — schema, auth, headers, and limits.
Knowledge Base (RAG)
Add documents the AI retrieves from — chunking, embedding, tags, and retrieval.
Naming. The Console shows Endpoints and Knowledge Base; the API and MCP call them Tools and RAG Documents. Same things — see the vocabulary map.