# What is a second brain for AI agents? A plain guide | Team0 > Why AI agents need a second brain, what it holds, how agents read and save to it, three ways people build one, and what to look for in any of them. Agent-readable version of https://team0.ai/second-brain. The page for people is at that address; this file carries the same content without layout. --- # What is a second brain for AI agents? One place your AI agents read about your work and your life, so you stop explaining the same things to each of them. ## Why agents need one. Agents are sharp. What they lack is your context. Each agent knows only what you told it : Your people, your projects, what you decided last week: an agent has none of it until you explain. With more agents, you become the messenger : Tell one, then repeat it to the next. Paste, explain, repeat. Your work keeps moving : Meetings move, plans change, people leave. What an agent heard last week may not be true today. Every new agent starts over : Add a new agent, or switch to a better one, and you explain everything again. ## What a second brain holds. The things you would otherwise explain at the start of every chat. People : Who they are, where they work, how you know them. Work : Projects, tasks, decisions, and what is still open. You : Your role, your preferences, how you like to work. What changed : The latest version of each thing, and what it replaced. ## How agents use it. Some agents do this on every turn by themselves. Others do it when their instructions tell them to. 1. 01ReadBefore it answers, the agent reads what is relevant to your request. 2. 02AnswerIt answers with your people, projects and decisions in mind. 3. 03SaveAfter it answers, what is new or changed is saved for next time. ## Three ways people build one. Each has a place. Here is what each is good at, and what to watch for. - 01 Each agent’s own memory Built into ChatGPT, Claude, Claude Code, Codex, OpenClaw and others. Good for : Nothing to set up. Watch for : Each agent remembers only its own chats. They do not share with each other. - 02 Markdown files A folder of notes (about-me, people, projects, tasks) and an instruction for your agents to read and update them. Good for : Plain text you can open and edit. Watch for : Every update is the agent’s judgement or your time. A new agent can use it only once it can open the folder and has your instructions. [Where markdown second brains break](https://team0.ai/markdown-second-brain) - 03This is Team0 A second brain that builds itself A service that learns from the work you connect and your agents’ conversations, keeps itself current, and serves every agent you connect. Good for : No upkeep. Every agent reads the same thing, and a new one starts with your context. Watch for : Some agents read and save on every turn by themselves. Others do it when they follow their instructions, and that is still their judgement. [See how Team0 works](https://team0.ai/how-it-works) ## What to look for in any second brain. Six questions worth asking, whichever way you go. 1. 01 Where it learns from Only what you type, or also your email, calendar, meetings and the tools you use? 2. 02 Who keeps it up to date You, your agent’s judgement in the middle of a task, or the system itself? 3. 03 What happens when something changes Is the old version overwritten, kept in history, or left next to the new one? 4. 04 Heard versus decided Can it tell something mentioned once from something actually agreed? 5. 05 Where each line came from Can you see the source of what it holds, and correct it in one place? 6. 06 New and switched agents Does a new agent, or the one you switch to, start with your context? Can you limit what each one sees? ## Words you will hear. Context : What an agent knows at the moment it answers you. Memory : What an agent keeps from one chat to the next. Second brain : One place that holds your context, so every agent can read it. MCP : A common way to connect an agent to a tool or service, such as a second brain. ## Keep learning. - [Markdown second brain vs Team0Where files quietly break, shown with one email](https://team0.ai/markdown-second-brain) - [Memory vs understandingWhat a memory layer does, and what understanding adds](https://team0.ai/memory-vs-understanding) - [How agents connectWhich agents read and save on their own](https://team0.ai/works-with) - [Follow one factOne email, from your inbox to every agent](https://team0.ai/world-model/architecture) - [A second brain you have to feed isn’t oneWhy we started building Team0](https://team0.ai/blog/building-team0/a-second-brain-you-have-to-feed-isnt-one) ## Your agents will change. Your understanding shouldn’t. [Markdown second brain vs Team0](https://team0.ai/markdown-second-brain) Invite-only private beta.