How to configure your agent's model and behavior

Agent Settings controls how your agent works: which AI model it uses, whether it understands images and voice notes, how long its answers can be, how much of the conversation it remembers, and how long it waits before replying.

Step 1: Open Agent Settings

In the left menu, open Agents and select your agent. Open Main, and in Prompt, click Settings above the prompt’s text field. The Agent Settings panel opens. Scroll inside it to see every setting.

Step 2: Set the timezone and the model

• Timezone: the time your agent uses, for example to book appointments in the right hour.

• Model: the AI model that powers your agent. Models differ in capabilities, speed and cost. The list shows only the models your agent can use.

Step 3: Turn on image and audio recognition

In Capabilities, check the features you want and choose a model for each one:

• Audio Recognition: turns your customers' voice notes into text so the agent can answer them.

• Image Recognition: lets the agent understand the images your customers send.

Step 4: Adjust how it answers

Move each slider:

• Temperature (0–2): determines how varied the answers are. Low values (0–0.5) make the agent more focused and predictable; high values (0.7–2), more creative. Default: 0.7.

• Max Tokens (256–8192): delimits the longest answer the agent can write. 1 token is about 4 characters or 0.75 words. Default: 2048.

• Memory Limit (20–100): defines how many previous messages the agent remembers. Higher values give more context but cost more; lower values are faster, but the agent may forget earlier parts of the conversation. Default: 100.

• Response Delay Min and Max (S) (0–180): sets how many seconds the agent waits before replying, picked between the minimum and the maximum, for a more natural pace. The maximum must be equal to or greater than the minimum. Default: 0, it replies right away.

Step 5: Save your changes

Click Update below the prompt. The same button saves your prompt and your settings, so your prompt needs at least 100 characters to persist the settings.

Shortcut: change the model while testing

In Test Agent, you can also change the model and its Max Tokens and Temperature from the message box: click the model name or the settings icon. These changes apply to the agent from the next message.

Done

Your agent works with the settings you selected. Test it after each change, and go back to the default values if the answers get worse.