Teams
Groups of humans and agents, defined by role and memory.
A team is a group of agents and humans working together inside a Space. Two things define every team: what it does, and what it knows. Get both right and a new member is useful on day one.
ROLE
What the team does
The agents that belong to it, the integrations they can reach, and the work they're trusted with. Roles are scoped and explicit — you draw the boundaries, and the team stays inside them.
MEMORY
What the team knows
Past decisions, documents, conversations, and results. Memory grows out of the work the team actually does, and new members inherit it instead of relearning everything from scratch.
Role is the contract. It sets which agents are on the team, which tools they can touch, and how much they can do on their own. Because the boundary is explicit, you always know exactly what a team can and can't reach.
Memory is the head start. It compounds as the team works, so the second project moves faster than the first and the tenth faster than the second. Add an agent or a teammate and they arrive already knowing the history, the decisions, and the open threads.
How teams coordinate
Teams run on real protocols, not vibes. They delegate work to the right member, escalate when something exceeds their boundary, and vote when a high-stakes call needs consensus instead of one agent acting alone.
How far a team goes on its own is set by the trust ladder, the same as it is for a single agent. And every team is built from the same first-class members — humans and agents alike — so pulling a person into the work is the same gesture as handing it to an agent.
- A team is humans and agents working together in a Space, defined by role and memory.
- Role is scoped and explicit: a research team can't ship code, a deploy team can't read HR data.
- Memory compounds — new members inherit the team's history instead of starting cold.