If you’re experimenting with AWS Strands, you’ll probably hit the same question I did early on:
“How do I make my agents remember things?”
In Part 2 of my Strands series, I dive into sessions and state management, basically how to give your agents memory and context across multiple interactions.
Here’s what I cover:
- The difference between a basic ReACT agent and a stateful agent
- How session IDs, state objects, and lifecycle events work in Strands
- What’s actually stored inside a session (inputs, outputs, metadata, etc.)
- Available storage backends like InMemoryStore and RedisStore
- A complete coding example showing how to persist and inspect session state
If you’ve played around with frameworks like Google ADK or LangGraph, this one feels similar but more AWS-native and modular. Here's the Full Tutorial:
Also, You can find all code snippets here: Github Repo
Would love feedback from anyone already experimenting with Strands, especially if you’ve tried persisting session data across agents or runners.