Shared Memory
A common store that several agents can all read from and write to.
Think of it like
A shared kitchen fridge — everyone in the house can take from it and add to it.
Example
A team of agents keeps its running findings in one shared store, so a worker that starts later can see what earlier ones already learned.
How it actually works
Shared memory is how multi-agent systems avoid re-doing each other’s work, but it brings all the classic problems of shared state: race conditions, stale reads, and one agent clobbering another’s entry. You trade the simplicity of isolated agents for the efficiency of pooled knowledge. In practice it needs conventions — who writes where, how conflicts resolve — or it turns into a mess fast.
For product teams
Pooled memory makes agent teams more efficient but adds the coordination costs of anything shared.
For engineers
A concurrently accessed store across agents, subject to consistency and contention concerns.
Related
- Blackboard — A structured way to use shared memory.
- Multi-Agent — The multi-agent setups that rely on it.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome