Solving LLM / AI Memory Loss: A Guide to Checkpointers in LangGraph

LANGGRAPH AGENTS DON’t remember by default. Let’s fix that, ? live. ➜ Why attend? See it break, then fix it. We’ll start with a plain LANGGRAPH AGENT, watch it lose state, and patch the hole with a checkpointer. Understand the moving parts. What gets stored, when, and how to rewind or branch a run. No magic, just code. Leave with sample snippets you can paste into your own repo tonight. ➜ What we’ll cover (live coding) How LANGGRAPH’s checkpointer API works under the hood Patterns for persisting and replaying state (Git-style history) Debugging a multi-step tool chain without rerunning from scratch ➜ Facilitator Anil Dukkipatty — AI educator who’s trained 800 + devs on LANGGRAPH, from weekend hackers to startup teams. ➜ Who should join? Developers shipping (or about to ship) LANGGRAPH WORKFLOWS AND TIRED OF STATELESS AGENTS. ➜ Tooling we’ll use We’ll use Convo (https://www.npmjs.com/package/convo-sdk [https://www.npmjs.com/package/convo-sdk]) as a drop-in checkpointer. Keep or swap it as you like—the patterns apply regardless. Built by Builders, Backed by the Best We’ve hosted engineers, founders, and researchers from some of the world’s most forward-thinking companies. Here’s a look at who’s been in the room.