Building Knowledge Graphs

This is a hands-on workshop where you'll walk away with a custom knowledge graph.

You'll own an end-to-end pipeline for generating a knowledge graph and executing graph queries, all packaged within a single Antfly environment.

What we cover#

• Extracting entities and relationships from documents with a local model
• Storing them as first-class relationships alongside the documents themselves
• Querying across the graph and the text in one request
• Keeping the graph current as documents change

Who it is for#

Anyone who has wanted to use the relationships in their data to answer questions that span documents. No graph database experience needed.

What to bring#

A laptop.

And while none of the following are strict requirements, they represent the happy path for the workshop. You can get a head start by having the following things ready for your environment:

• A MacBook Pro: While not strictly required, the environments we'll be setting up in the workshop will assume a macOS environment.
• Coding Agent: We recommend OpenAI Codex, as we'll be deploying our agent with ChatGPT sites. (You can find similar tools with Claude Code Artifacts, or DIY deployments with Vercel, but the workshop will walk through things with Codex.)
• OpenAI API Billing: Having a frontier model provider API with a usable API key will dive right in to the workshop.
• SearchAF: This is Antfly's macOS semantic search app, and will help to get an Antfly environment up and running.
• A Folder of Documents: This will be our "knowledge base": all of the documents you want to ground your agent with. This can be a Google Drive folder, a folder on your desktop. The format of the documents in the knowledge base could be just about anything — markdown, Google/Office docs, HTML, PDF, MP3, etc.