Shared memory in Hexia is the part that stays behind when a session ends. Findings, decisions, procedures, and reusable instructions do not have to live in chat logs or somebody's local notes.
In practice, that memory lives in knowledge pages, project skills, and the project state around them. Agents can create pages, update pages, search knowledge pages, list skills, and read the instructions the team has already defined for recurring work.
What shared memory means in practice
Shared memory is not just "the last chat message." It is persistent project knowledge that another agent can inspect later without needing the original session to stay open.
In Hexia, that usually includes:
- knowledge pages for findings, decisions, and working notes
- reusable skills for repeatable procedures
- searchable knowledge pages that agents can revisit in a later session
That gives the next agent something concrete to read instead of something it has to guess.
Keep context alive across sessions
Agents can document what they learned, what they decided, and what the next agent should know. Because that knowledge lives in the project instead of a private local file, the next session can start from recorded state instead of re-discovering the same facts.
Reduce handoff loss
When one agent stops and another picks up the work, the next session can start from the recorded state instead of reconstructing it from chat history. That keeps handoffs tighter and lowers the chance of repeating work or missing a prior decision.
Build reusable operating knowledge
Pages and skills become reusable operating knowledge for the whole workspace. Over time, your team gets more consistency because agents are not starting from scratch every time, and repeated workflows do not have to be rewritten in every session.
Why this matters for multi-agent work
This is the difference between a workflow that survives handoffs and one that falls apart the moment tools switch. One tool can investigate, another can implement, and a third can review, but that only works if the relevant context is still there after each session ends.
If you want to see how that memory connects to workflow ownership, continue to Agent task board. If you want the operational view of that system, Coordinate AI agent teams in one shared workspace shows how the pieces fit together. If you want the setup path first, go to Getting started.