Agents and Threads

Two ideas organize your work in Trezable AI: agents and threads.

Agents

An agent is a named assistant with its own model and skill set. You can create several agents — a general assistant, a deep researcher, a writer — and switch between them depending on the task.

• Create an agent — give it a name and choose the model it uses.
• Select skills — choose the capabilities the agent should have.
• Update an agent — change its model or skills at any time.
• Switch agents — pick the agent you want from the conversation interface.

Each agent is independent: configuring one does not change the others.

Threads

A thread is a single conversation — its own history, messages, artifacts, and state. The sidebar lists your threads so you can return to any of them.

• New Thread — start fresh; the new conversation does not inherit the previous one’s history.
• Rename a thread to find it easily later.
• Continue a thread — pick it from the list and keep going.

How they combine

Agents define what the assistant can do; threads define what the conversation is. You can start many threads with the same agent, and each thread stays exactly as you left it. Switching agents does not disturb your threads, and switching threads does not change your agent.

A typical workflow

1. Create a research agent for your project.
2. Start a thread and ask it to research a topic.
3. Start a second thread with the same agent for a different topic.
4. Return to either thread later — history, artifacts, and state are intact.