# Multi-Agent Architecture

> The ideas behind The One: what makes an Agent and keeps it alive, when another Agent pays off, and how teams are written down, grow, work together and improve.

This section explains why The One is built the way it is. You don't need to read it to use The One; come here when you want the reasoning behind the main assistant, working groups and evolution.

## Read it in this order [#read-it-in-this-order]

1. [Single Agent](./single-agent): an Agent is determined by its model, its tools and its context.
   * [Context and meditation](./single-agent/context): what it sees, and what happens when its context fills up.
   * [ReAct and turns](./single-agent/react): how it works step by step, and when a turn ends.
   * [Agents that stay alive](./single-agent/life): it is still there after sleeping, errors and app restarts.
2. [From Single Agent to Multi-Agent](./evolution): when another Agent is worth adding, and when it is not.
3. [A-Genetic Engineering](./a-genetic-engineering): how a team is written down in full, so you can reuse it and pass it on.
4. [Prolifera Engineering](./prolifera-engineering): how a team grows while it works.
5. [Collaboration: in a group, between groups, and with you](./collaboration): how work passes between Agents, between working groups, and between them and you.
6. [Self-Evolution](./self-evolution): how a Space gets better at its own work from evidence.
7. [No approval step](./record-and-undo): Agents act on their own; safety comes from the record, undo and stopping.

To use these ideas in the app, start with the [Quick Start](../start/quick-start) and [The main assistant](../the-one-app/main-assistant).

Source: https://the-zeroth.com/en/docs/multi-agent-architecture