teta.so
Sign in

Parallel Execution

When a task is large enough, the agent can suggest splitting the work across multiple chat tabs. Each tab runs its own independent agent, so you can have two or more agents working on different parts of your project at the same time.

How it works

  1. You send a complex request to the agent in your current chat.
  2. The agent analyzes the task and identifies parts that can be worked on independently.
  3. Instead of doing everything itself, the agent gives you a ready-to-send prompt for a second task.
  4. You open a new chat tab and paste that prompt. A second agent starts working on it.
  5. Meanwhile, the first agent continues with its own part of the work.

The key point: you are in control of when and how to split the work. The agent suggests the split and provides the prompt, but you decide whether to create the new chat.

Why split work across agents

Each agent in a chat tab has its own context and focus. Splitting work gives you two advantages:

  • Speed — Two agents working in parallel finish faster than one agent doing everything sequentially.
  • Specialization — The first agent can focus on reviewing and coordinating while the second agent focuses on implementation. This produces higher-quality results than a single agent trying to do both.

Code review across agents

One powerful pattern is using the first agent as a reviewer for the second agent's work:

  1. Agent A (your main chat) gives you a prompt to send to a new chat.
  2. Agent B (the new chat) implements the feature.
  3. You go back to Agent A and ask it to review the code that Agent B wrote.
  4. Agent A reads the files, reviews the implementation, and gives you feedback.
  5. You send that feedback to Agent B to apply the fixes.

This creates a review loop where one agent writes and another reviews — similar to how a team of developers works, with a natural quality gate between writing code and merging it.

When to use it

Parallel execution works best when:

  • The task has clearly independent parts (e.g., building a header component and an API route).
  • You want a code review step without slowing down implementation.
  • You are working on a large feature and want to make progress on multiple fronts.

For smaller tasks or tasks where every step depends on the previous one, a single chat is simpler and more effective.