Skip to main content
Intelligence settings control the model your Agent uses and whether it spends additional time reasoning through difficult requests.

​
Choose a model

Open Edit Agent and select the Intelligence section. Choose from the models available to your account. The best model depends on the job:
  • use a fast model for straightforward routing, scheduling, and frequently repeated questions
  • use a stronger model when conversations involve complex policies, troubleshooting, or several pieces of information
  • test response speed as well as answer quality, especially for phone calls
Available models can differ by account and plan.

​
Use reasoning intentionally

Reasoning can help with multi-step decisions, comparisons, and ambiguous requests, but it can also increase response time. Enable it where the work benefits from additional analysis rather than applying it to every simple exchange.

​
Pause & Think

Pause & Think gives the Agent a private reasoning step during a conversation. It is useful when one particular decision needs more analysis without making every response slower. Test the staged Agent with realistic examples before publishing, including the hardest decision it is expected to make.
Configuring models programmatically? See Models and reasoning.