One email to Kate, start to finish: how her brain checks it and keeps it true, how it reaches the company brain, and how your Codex gets it, only where you allow.
Every AI agent you use only knows what you’ve told it.
Not your work, your people, or the meeting that moved an hour ago. Team0 gives every agent one shared understanding. To see how it works, follow Jordan’s email down the middle: Kate’s brain, the company brain, your Codex.
01SourcesCan we move the pilot review to Thursday?
It starts in the tools you already use.
Team0 syncs your sources in the background. Nothing to forward, nothing to tag. Gmail is read-only.
Gmail
Calendar
Documents
Notion
HubSpot
GitHub
monday.com
Stripe
QuickBooks
Slack
and more
Jira
Meetings
Email from Jordan to Kate
“Can we move the pilot review to Thursday?”
Tue 09:12 · this is only the ask
Kate’s calendar
Fri 11:00 Thu 14:00
Tue 09:40 · Pilot review moved
Two signals about one meeting, both in Kate’s brain. The difference between them matters later.
Every input becomes a background job.
It lands in a queue, deduplicated, so nothing is processed twice. Queued by urgency, worked on your server.
urgent
soon
routine
background
workersyour server
Accepted input
Encrypted, deduplicated.
Your words, agents’ actions
Agents’ reported actions are kept as heard.
Questions save nothing
A question is never a fact.
Originals kept as long as the company chooses
90 days by default. What was learned from them stays.
What a source supports matters.
Team0 keeps the source, attribution and uncertainty visible. Here is what that means for the answer an agent receives.
Attribution:Who made the statement and which source supports it
Uncertainty:A request or a report remains distinguishable from a confirmed change
History:An earlier plan remains available when the current one changes
Human control:You can inspect and correct what Team0 understood
Heard is not confirmed
Pilot review: Thursday 14:00
confirmed
Kate’s calendar moved the meeting at Tue 09:40. That is the record, so the fact is confirmed. Jordan’s email stays attached as the reason it moved.
An answer with context.
A fact is more than a sentence. It can explain where it came from, what remains uncertain and which earlier plan changed. Nothing is silently overwritten.
Pilot review: Thursday 14:00confirmed
Current plan
The pilot review is Thursday 14:00
what the agent can use for this example
People and work
Jordan · Northpeak · Pilot review
the context around the answer
Source
Kate’s calendar changed on Tuesday
why the answer changed
Earlier plan
Friday 11:00
kept as history
Your control
Inspect or correct the understanding
when something needs fixing
See the plan change over time
Drag through the illustrative week to see the request, the confirmed change and the meeting in the past.
Example: Jordan's email arrives Tue 09:12. Until Tue 09:41 the record is still Friday 11:00, and Jordan's ask is kept as heard. The calendar move was said at Tue 09:40 and learned at Tue 09:41; from then the fact is Thursday 14:00, confirmed, with Friday 11:00 kept in history. The date it is about is Thu 14:00.
Fri 11:00Thu 14:00A newer value takes over. The calendar moved it at Tue 09:40; the old value stays in history.
Facts link into a picture, and it keeps itself current.
One private brain per person, plus the company brain. People, companies and moments link into storylines, and every night the whole picture is brought up to date.
Records are the circles.Facts are the links and statements between them.
People and companies
The context around the work you do together
Meetings and conversations
Sources you can inspect for the details
Decisions and promises
What changed, what is open and what needs attention
History and corrections
A way to understand and fix an answer
Keeps itself current
A maintained product needs ongoing care as sources, models and agent hosts change.
Source health
Know whether connected information is still arriving
Quality
Inspect answers and correct mistakes
Retention
Keep information under the chosen policy
Backups
Recovery is part of operating the system
Updates
Maintain compatibility as providers and hosts change
Costs
See what the product’s model work costs
Kate shares her work facts with the company.
Each person has sharing controls. Coworkers cannot open your private brain through Team0.
Kate shares with the company
Pilot review: Thursday 14:00
her choice · her name goes with it · private stays home
Your sharing choice
Sharing starts off; your explicit choice takes precedence
Attribution stays visible
Shared knowledge names who shared it
Separate source access
Sharing a fact does not open a coworker’s original sources
Withdraw a shared item
Remove it from company use and retained shared copies
Sensitive information
Classification helps hold it back, but models can miss
Your private brain
Coworkers and admins cannot open it through Team0
Coworker messaging
A separate opt-in; message bodies are not screened
Know the limits
Server operators can read the database; delivered information cannot be unread
Company knowledge reaches the agents allowed to use it.
Company sources and permitted contributions provide context beyond what one person told one agent.
Company sources
Learn from the work sources the company connects
Shared knowledge
Use permitted contributions with their attribution
Relevant context
Help agents work from current company information
Access boundaries
What an agent receives depends on its permissions
Uncertainty visible
Unconfirmed claims and disagreements remain distinguishable
People in control
Inspect, correct and remove shared information
Agent connections
Company knowledge can be disabled for an individual agent
Team administration
Invite people and pause or revoke their access
Later, your Codex asks.
You never saw Jordan’s email or Kate’s calendar, and your Codex was never told about the pilot review. But Northpeak is already in your brain.
Your Codex asks Team0:
“When is the pilot review with Jordan?”
One read, shaped by the question.
An agent receives context for its task, within its access, with sources and uncertainty it can inspect.
Relevant context:Information that helps with the task the agent is doing
Current or uncertain:A clear distinction between the plan and unresolved claims
Sources and history:Attribution and the earlier values that matter to the answer
Access:Personal and company information within the agent’s permissions
Your decision:Questions the agent can raise when a person needs to decide
Thu 14:00 · confirmedshared by Katepacket scoped to Codex
Controls around each agent connection.
Each agent has its own connection. Explore the controls available to its owner.
identity: Each connection belongs to an authenticated agent and person.
access: Choose what understanding this agent may read.
company: Control company knowledge separately for this agent.
sources: Choose whether the agent can open the full sources it is allowed to use.
activity: Inspect what the agent asked and received.
retention: Choose how long the kept read content remains available.
revoke: Stop this connection without revoking your other agents.
Company.Control company knowledge separately for this agent.
Codex
readssaves
Codex gets “Thursday 14:00, confirmed. Kate shared it with the company.” What it learns can be saved back.
Tools through the door: read, write back, and message your own agents, or a coworker’s if they allow it.
Stop saving, or stop all access, per agent, any time.
The answer carries its proof.
Not just a time. The agent can say where it came from: confirmed, not merely asked for, and shared by Kate.
Your Codex asks:
“When is the pilot review with Jordan?”
Team0 answers:
“Thursday at 14:00, confirmed. Kate shared it with the company.”
confirmedshared by Katesaid Tue 09:40Fri 11:00 in history
Every agent reads the same understanding.
And can write back to it. Each one sees only what you allow. Point at an agent to see how it connects.
Every turn
A plugin reads before each turn and saves after it.
Claude Code
Codex
OpenClaw
Hermes
On demand
MCP plus standing instructions that tell it when to read. What they relay is kept as heard.
Any agent that connects over MCP
Codex reads Team0 before every turn and saves after it, so it starts each turn knowing the pilot review is Thursday.
They talk to each other
Your agents message each other through the same door, and a coworker’s agents too, if they allow it.
They take your meeting notes
A notetaker named after the agent you pick joins your calls and feeds meetings straight to your brain.
Your console
Brain · Agent reads · Setup · Company
Fix a fact once and every agent gets the fix
Pause a source, set never-learn rules, see the exact text each agent got.
What is kept, and for how long.
Originals don’t stay forever. Learned facts stay, with their history; the originals around them are kept as long as the company chooses, 90 days by default, then erased by a daily job.
Full read results: 1 day by default, up to 30, set per person
The read log’s content: 24 hours to 30 days, set per person · its record stays 30 days
Agent conversation words: 90 days by default, 30–365, set by the company
Other agent and API inputs, encrypted: 90 days by default, 30–365, set by the company
Meeting recordings: 7 days · the transcript and summary stay
Facts learned, with their history: kept until withdrawn or deleted · never silently overwritten
1 day7 days30 days90 dayskept
Full read results
1 day by default, up to 30, set per person
The read log’s content
24 hours to 30 days, set per person · its record stays 30 days
Agent conversation words
90 days by default, 30–365, set by the company
Other agent and API inputs, encrypted
90 days by default, 30–365, set by the company
Meeting recordings
7 days · the transcript and summary stay
Facts learned, with their history
kept until withdrawn or deleted · never silently overwritten
A maintained company install, with updates and support.
Signed nightly updates
Verified releases and recovery checks help maintain the install.
Your choice of models
Each provider’s own keys, or your own Amazon Bedrock.
Daily health checks
Every model, the address, the last update. Counts go to Team0 only if you say yes.
An agentic brain, from source to answer.
Follow one meeting change through a private brain, the company brain and an agent’s answer. Explore the sources, changing facts, evidence, permissions and controls around them.