← All guidesFollow the thread

Use Brain with your AI agent and connection graph

Using several agents—Codex, Claude, Muse—or switching between AI accounts? Keep your work and its context in Brain so you can carry the plan from one conversation to the next. Start with a guided setup, then use your connected agents to capture tasks, triage commitments, connect related work, and pick up where you left off.

Read as Markdown ↗

Set up shared context for your agents

Give your different agents a shared place to pick up the work. Use Brain for the project brief, decisions, people involved, and next actions you want to carry between tools and accounts.

Copy this prompt to a connected agent. It will help you organize one current project and save the context your other agents need to work on it.

You can start with an empty Brain account. The agent will help you create the structure you need, or reuse relevant areas and items if you have already saved some work.

Try asking your agent

Use Brain to organize my work and keep shared context across my agents and AI accounts. Walk me through the setup one question at a time. Ask which agents I use, what work I move between them, and what each needs to know. Check what is saved in my account. If I have no work saved yet, start from the work I describe and help me create the first project, tasks, notes, and relevant contacts. If I already have work saved, read it, reuse suitable items, and help me fill in the gaps. If I have no areas beyond the built-in Inbox, help me choose an area to create; otherwise, reuse the area that fits. Ask about my responsibilities and what I am working on, then help me choose one real outcome to start with. Suggest an area for that responsibility and a project within it. Ask about the people involved, useful background, decisions already made, and next actions. Propose the missing areas, project, contacts, notes, and tasks using the information I provide. Put project tasks in the project and standalone tasks in the appropriate area. Suggest Related to connections that explain the work and Depends on connections between tasks that need to happen in sequence. Show me the setup, then create or update the items once we agree. Finish by showing me what you saved and giving me a short instruction I can use with each connected agent to read the current context before starting and save decisions and next actions when we finish.

Not connected to an agent yet? Check out our documentation to connect ChatGPT or Claude to Brain.

Give your agents the same starting point

Your agents and AI accounts can change while your work stays in one place. Connect each supported agent to the same Brain account so it can read and update your saved areas, projects, tasks, notes, and contacts. One can help you plan the work; another can act on the plan you saved.

If a useful conversation happened elsewhere, bring its summary to a connected agent and ask it to save the decisions and next actions in Brain. Keep the project brief and key constraints there too. At the start of the next session, ask the agent to read the current plan; at the end, ask it to save what changed.

Choose the connection that explains the work

Connection	Use it for	What it gives your agent
Task home or note attachment	Keeping work and its supporting information together	A place to find the next actions and supporting information for a project
Related to	Connecting people and context across your work	A way to find the people, notes, and other work relevant to an item
Depends on	Making the sequence of tasks clear	The prerequisite to check before starting the next task

A task can stay in its project while also being connected to a client contact or a useful note. You can reach the same work from whichever item you start with.

A dependency shows what should happen first. You can still complete tasks in any order, so check the prerequisite before marking a dependent task done.

Turn your notes into tasks and context

Paste a meeting summary, working notes, or a list of loose ends into your agent conversation. Ask it to separate commitments from background information, then save both where you will find them.

Try asking your agent

Help me turn the notes I paste next into useful items in Brain. Read them for commitments, next actions, decisions, and questions. Check for relevant areas, projects, contacts, and tasks already saved. If I only have Inbox, ask what responsibility or outcome these notes relate to and help me choose an area or project to create. Reuse matching items if they exist, and propose any missing areas, projects, or contacts. Suggest tasks for the work I need to do and notes for the context worth keeping. Ask about unclear commitments and missing planned dates or deadlines. Show me the proposed items and where each belongs, then create or update them once we agree. Connect the supporting notes and people, and show me what you saved.

Triage what needs your attention

With a base setup in place, ask an agent to help make sense of new Inbox items and unfinished work. It can suggest next actions, put tasks in the appropriate areas or projects, and identify commitments that need a decision.

Try asking your agent

Help me triage my Brain workspace. Check what is saved in Inbox and any areas, projects, tasks, notes, and connections. If there is no work saved yet, help me capture my current commitments and set up a few areas first. If tasks are only in Inbox, ask about my responsibilities and propose areas or projects for them. If I already have a structure, reuse what fits. Identify approaching or overdue deadlines, tasks waiting on someone or another action, and projects without a clear next step. Ask about unclear items one at a time. Suggest a concrete next action and a home for each task. Show me the proposed changes and help me choose what to work on next. Once we agree, create any missing areas or projects, save or move the tasks, and record useful decisions in the relevant notes. Summarize what changed and what still needs my input.

Find useful connections in your existing work

Once you have saved tasks, notes, and contacts, ask an agent to find relationships that will help you act. A conversation note can explain a follow-up task; a contact can lead you to all the work involving that person.

Try asking your agent

Help me find useful connections in Brain. Ask which area or project to review, then read its tasks, notes, contacts, and existing connections. Look for notes that explain a task or decision, people involved in the work, related projects, and tasks that need another action to happen first. Show me the connections you recommend and the saved information that supports each one. Once we agree, add the missing Related to and Depends on connections, keeping each task in its current home unless we agree to move it. Show me what you connected.

Open Graph view from the project’s Connections section, or Explore graph from a contact’s page. Select an item to follow its connections; use Back to retrace your path and Open item to read or edit it.

Not connected to an agent yet? Check out our documentation to connect ChatGPT or Claude to Brain.

Use the connections to ask better questions

Prepare for a meeting: start with the contact and gather the notes, commitments, and projects you need to discuss.
Brief an agent on a project: bring the outcome, scope, decisions, people, and next actions into one conversation.
See what needs to happen first: follow task dependencies before choosing the next piece of work.
Follow a decision into action: connect a decision note to the tasks that carry it out.
Spot gaps: ask which projects lack a next action or which client conversations still need a follow-up.

Try asking your agent

Help me prepare a project briefing. Ask which project I want to review, then read it along with its tasks, notes, and connected contacts in Brain. Summarize the agreed scope, decisions, next actions, and open questions. Show the items you used.

Save the useful result of a conversation

An agent session may produce several options and a final decision. Save the decision, its reasoning, and the action that follows. That gives the next agent a clear place to begin.

Ask the agent to find the existing project and contacts before it saves changes. Keep the updates on those items so the project history stays together.

Try asking your agent

For the project we just discussed, save a note summarizing our decision and why. Add the agreed follow-up task to the project and connect it to the people involved. Use the existing project and contacts, ask if anything is unclear, and show me what you saved.

Hand work from one agent to another

Treat the project notes as a handoff brief. Include the goal, constraints, current decisions, unanswered questions, and the next action. The next agent can read that brief and inspect the connected items for detail.

Connect each assistant separately to the same Brain account. They can share the context saved in Brain; save the parts of a conversation you want another agent to use.

When two agents are revising the same plan, let one finish and save its changes before the other picks it up. Ask the next agent to read the current notes before making edits.

Try asking your agent

Help me pick up a project another agent worked on. Ask which project, then read its current notes, contacts, and tasks in Brain. Summarize where the work stands, what has been decided, and the next action before we continue.

Keep the context worth reading

Save enough detail to explain a decision and support the next action. Record what was decided and why, then connect the note to the work it affects so an agent can understand the reason for a task.

At a project review, update the brief to reflect the current plan. Ask your agent to distinguish what has been agreed from what it recommends. That keeps future conversations grounded in the work you have actually committed to.