Agent-level configuration
Selecting and configuring AI models for individual agents
Each agent can be configured to use either Kapso AI credits or your own API keys. This gives you flexibility to use different approaches for different agents within the same project.
Configuration process
Go to agent settings
Access the settings area for the specific agent you want to configure.
Choose billing method
Select how you want to power this agent’s AI:
- Use Kapso credits: Simple, no API keys needed
- Use your own API keys: Direct billing from providers
Configure based on your choice
If using Kapso credits:
- Adjust temperature, max tokens, and reasoning effort
- That’s it! Claude Sonnet 4 is automatically selected
If using your own keys:
- Select your provider (OpenAI, Anthropic, etc.)
- Choose the specific model
- Configure parameters
Save configuration
Save your agent’s AI model settings.
Model parameters
Regardless of billing method, you can customize these parameters:
-
Temperature (0.0 - 1.0): Controls creativity vs. predictability
- Lower values (0.0 - 0.3): More focused, deterministic responses
- Higher values (0.7 - 1.0): More creative, varied responses
-
Max tokens: Maximum length of generated responses
- Prevents runaway token usage
- Ensures responses fit within message limits
-
Reasoning effort (for supported models): How much computational effort to use
- Low: Quick responses for simple tasks
- Medium: Balanced performance (default)
- High: Maximum effort for complex reasoning