How Team0 keeps your agents in the know.
It learns from the work your team already does, keeps it current, and gives each agent only what it’s allowed to see.
We set it up. It keeps itself current.
We install it. From then on it maintains itself, learning from the work your team already does.
- 1We install itIn your own AWS or Google Cloud account, or on a server of its own that we run for you.
- 2People joinEach person accepts an invitation and connects their Google account, then any other tools they use.
- 3Agents connectEach person connects the agents they use. Each agent gets its own access.
- 4Your agents are in the knowEach agent starts from its person’s brain and what colleagues chose to share.
Claude Code, Codex, OpenClaw and Hermes read and save on every turn, through a plugin. Other agents connect over MCP, with standing instructions that tell them when to read.
A brain for each person. One for the company.
What one person learns can reach a colleague’s agent, but only what they chose to share.
Your brain is yours
Built from your own work. Not the admin’s, not your coworkers’.
You choose what crosses
Only work facts, each with your name on it. Withdraw any time and it’s gone within a minute.
The company has its own
Shared inbox, calendar, CRM, documents and Slack channels, plus what people share.
When someone leaves
What they shared stays with the company. Their private brain is erased.
Every answer shows where it came from.
Team0 keeps who said it, when, and what it replaced. Nothing is quietly overwritten.
Confirmed, heard, replaced, needs you.
A request is not a promise. Something said in passing is not a decision. Your agents are told which is which.
Backed by a reliable source, like a calendar invite. Treated as current.
Said in a meeting or chat, not settled. Passed on as reported, not as fact.
No longer current. Kept in history with what replaced it.
Team0 can’t decide this alone, so your agent asks you.
When Team0 isn’t sure, your agent asks you.
Which of two numbers holds, whether a promise is done, what a new codename means. The question rides in your agent’s next read, and your answer settles it for every agent.
When an agent asks, it gets only what it needs.
Agents don’t get a copy of everything. Each question gets a short answer, with sources.
- 1It works out what the question is aboutThe people, matters and dates it names, not everything Team0 knows.
- 2Access is checkedOnly what this agent may see is included.
- 3A short answer, with sourcesWhat is current, what changed, what is still open.
- 4On the recordEvery read is written down, so you can look it up later.
Agents add to it, too.
When an agent finishes a conversation, it can save what mattered. Your other agents get it.
Labelled by agent
You can always tell whether something came from Codex, Claude or your own email.
Checked like everything else
A decision made in a chat counts as a decision. A passing mention stays “heard”.
Yours to switch off
Stop an agent saving and it can still read. Stop all access and it’s cut off.
Agents talk to each other
Your agents can message each other, and a coworker’s agents when that coworker allows it.
What it doesn’t do.
Better you know the limits up front.
New work takes a little time
Email and calendar sync hourly, meetings when they end, other tools about daily. Agent turns arrive live.
It can be wrong
That’s why every statement shows its source, and you can correct it in one place.
It understands. Your agent acts.
The work is done by the agent you chose, with the permissions you gave it there.
Want the technical version?
The data model, identity, and how it stays current.
