What it takes to build agent memory for production
Redis' field engineering team will walk you through how agent memory fits into an agentic app, how to control what your agent remembers, and how to bring the right context back into the agent at runtime.
We’ll cover:
Session memory: Keep an ordered history of conversation events and retrieve the context your agent needs for the next turn.
Long-term memory extraction: Turn useful information from conversations into durable memories automatically, without building your own extraction pipeline.
Automatic summarization: Keep long conversations manageable by summarizing older events while preserving the most recent turns in full.
Custom memory types: Define the domain-specific information your agent should remember and guide how that information is extracted.
Memory policies & safety: Control retention with TTLs and guide extraction away from sensitive information your agent shouldn't keep.
Recall at runtime: Search and scope relevant long-term memories and provide them back to your agent when they're needed.
Who it's for
Built for the people writing the memory layer
Developers, architects, and platform engineers who are building agent memory into their systems. You want implementation guidance you can act on, not just validation that Redis performs well on a leaderboard.
This workshop is for you if:
Memory is what's blocking your agents from reaching production
Most of what you find is vendor marketing or a benchmark leaderboard
You're making the architecture decision now, without a framework to lean on
