Skip to main content
The final step in mastering Worlds is connecting an AI agent, such as Gemini, Claude, or ChatGPT, to your world memory.

​
Neuro-symbolic architecture

This integration bridges two paradigms:
  • Neural: The Large Language Model (LLM) that processes natural language.
  • Symbolic: The Worlds Graph that stores structured facts and follows logic.
By providing the LLM with tools to read and write to the graph, you enable the agent to maintain a persistent state and perform complex reasoning.

​
Using the AI SDK

The worlds-ai-sdk provides pre-built tools for frameworks like the Vercel AI SDK, making integration straightforward. 
import { generateText } from "ai";
import { openai } from "@ai-sdk/openai";
import { WorldsSdk } from "@wazoo/worlds-sdk";
import { createTools } from "@wazoo/worlds-ai-sdk";

const sdk = new WorldsSdk({
  baseUrl: "http://localhost:8000",
  apiKey: "your-api-key",
});

const { text } = await generateText({
  model: openai("gpt-4o"),
  tools: createTools({
    sdk,
    sources: [{ id: "my-knowledge-base" }],
  }),
  prompt: "Find all team members and summarize their roles.",
});

​
System flow

  1. User prompt: You ask the agent a question.
  2. Tool selection: The agent determines it needs information from its “World” and triggers a query tool.
  3. Execution: The worlds-ai-sdk executes a SPARQL query against the Worlds API.
  4. Inference: The graph returns structured data, which the agent uses to generate a factual response.
To learn more about connecting and monitoring these integrations in production, see our guides on Data Integration and Monitoring.

​
Next step

Proceed to Commencement to claim your certificate and explore the next stage of your journey.