A second brain with a body

Every other tool gives you a brain in a jar. You can ask it questions but it can't do anything. Fabric's brain has hands. It writes docs, runs agents, acts across your apps, and shares with clients. It doesn't just think. It works.

Every second brain tool follows the same model: you put things in, you take things out. Save a note, find it later. Clip an article, search for it when you need it. The second brain is a jar that holds your knowledge. Ask it a question and it answers. But the jar can't do anything with the knowledge it contains.

It can't write the meeting summary. It can't follow up with the client who hasn't reviewed the deliverables. It can't compile the weekly status report from the team's activity. It can't monitor your competitors' websites and alert you to changes. It can't take the research you've accumulated and produce a first draft of the document you need to write. It can't share a polished deliverable space with your client. It knows things but it can't do things.

A brain without a body is a reference library. Useful, but limited to answering questions you think to ask. A brain with a body is an assistant that acts on your behalf, using the knowledge it contains to do work that would otherwise consume your time.

What the body does

It writes

Self-writing documentation means the brain doesn't wait for you to ask for a summary. It generates decision records from Slack discussions. It writes meeting notes from recorded calls. It produces engineering wikis from GitHub activity. It creates project documentation from the team's ongoing work. The brain uses what it knows to produce the documents that would otherwise require someone to stop working and start writing.

It acts on schedule

AI agents run recurring tasks without prompting. A weekly summary of your business activity compiles every Monday morning from your email, calendar, and connected tools. Follow-up emails send automatically when a client hasn't responded to a proposal. Competitor monitoring runs on schedule and flags changes. Invoice reminders send at the intervals you configured. The brain doesn't wait for you to ask. It does the work on schedule, using the knowledge it has about your business, your clients, and your processes.

It works across your apps

Through connections and MCP, the brain extends into every tool you use. It reads from Slack, email, Google Drive, GitHub, CRM, and dozens more. It doesn't just search these tools. It acts across them: capturing context, synthesising information, and producing output that spans the boundaries between apps.

It shares and publishes

The brain can present its work to the outside world. Published pages create shareable deliverable spaces with link analytics (who viewed what) and password protection. Client handoffs that would take an hour to assemble are generated automatically from the project's accumulated context. The brain doesn't just know about the project. It can package the project for the client.

It collaborates

Through Fabric Tag, the brain is accessible in Slack. Tag it in a conversation and it searches the knowledge base, answers questions, and takes action from within the messaging tool the team already uses. No context switching. No app switching. The brain is where the conversation is.

From passive to active

The shift from a passive second brain (store and retrieve) to an active one (store, retrieve, write, act, share) changes what a second brain is for.

A passive second brain helps you remember. An active second brain helps you work. It handles the admin that consumes 30% of your week. It produces the documentation that nobody has time to write. It monitors the things you'd forget to monitor. It follows up on the things you'd forget to follow up on. It turns knowledge into action rather than storing knowledge as a passive archive.

The difference isn't incremental. A brain that can do things is a categorically different tool from a brain that can only answer questions. It's the difference between a reference book and an assistant. Both contain knowledge. One of them also has hands.

Frequently asked questions

What kinds of actions can the agents take? Agents handle recurring tasks: compiling summaries, sending follow-ups, monitoring sources, generating reports, creating task lists from meeting notes. The actions are configurable and run on schedule or in response to triggers.

Can agents send emails or messages on my behalf? Agents can draft communications for your review before sending, or send directly if you configure them to. Most users start with review-before-send and grant more autonomy as they build trust in the agent's output quality.

How does Fabric Tag work in Slack? Fabric Tag is a Slack integration. Tag @Fabric in any Slack channel or DM and it searches the knowledge base, answers questions, and can take actions, all from within Slack. The team doesn't need to leave the conversation to access the knowledge base.

Is the "body" just automation? It's automation informed by knowledge. A standard automation tool follows rules. An agent informed by a knowledge base makes decisions based on context. The follow-up email isn't a generic template. It's informed by the client's history, the project's status, and the specific deliverables that were shared. The knowledge makes the automation intelligent.

What about privacy? Can agents access everything? Agents operate within the access controls you configure. You specify which sources the agent can access, which actions it can take, and which content it can reference. The agent doesn't have blanket access to everything in your library unless you grant it.

How does this compare to tools like Zapier or Make? Zapier and Make automate workflows between apps based on triggers and rules. Fabric's agents do this with the addition of knowledge context: the agent can search your knowledge base, reason about the results, and take actions informed by accumulated context. The combination of automation and knowledge is what makes the output intelligent rather than mechanical.

Can I use agents without the second brain features? The agents work best with a rich knowledge base because the context makes their output more useful. You could use agents with minimal content, but the value grows proportionally with the knowledge the agent can draw on.

Is this safe? What if an agent does something wrong? Agents are configurable with review steps. For high-stakes actions (sending to clients, modifying data), configure the agent to draft for your review rather than act directly. For low-stakes actions (compiling summaries, flagging items), direct action is safe and saves time. The safety model is progressive trust: start cautious and grant more autonomy as you verify the agent's judgment.

What's the roadmap for agent capabilities? The trajectory is toward agents that can handle increasingly complex, multi-step tasks across connected apps: not just compiling a summary but identifying the most important items, drafting responses, and flagging decisions that need your attention. The more context the knowledge base accumulates, the more capable the agents become.

Related reading: Your second brain should think, AI for business owners, Agencies waste 30% of their time, How to build a company brain. Related pages: Agents, Self-writing docs, Fabric Tag, Teams, About.