Agent Presets

TL;DR:

• Presets define AI models, system prompts, temperature, and token limits.
• Create different presets for different roles (e.g., Support, Sales, Internal Helper).
• Check the scenario usage badge to see where a preset is already connected before modifying or deleting it.

Overview

Agent Presets define how your AI assistant behaves, what model it uses, and the rules it must follow. By creating reusable presets, you avoid writing the same system prompts or model configurations multiple times across different scenarios.

Location: Workspace Settings → AI Settings → AI Presets

Configuring a Preset

Field	What it does	Best Practice
Preset Name	An internal label for your team (e.g., FAQ Bot, Sales Qualifier).	Make it descriptive of the role.
Provider	The AI provider (e.g., OpenAI, Google Gemini).	Select based on speed/cost balance.
Model	The specific model version.	Use lighter models for fast triage, larger models for complex logic.
System Prompt	The core instructions and rules for the AI.	Be specific about tone, boundaries, and fallback behavior.
Temperature	Controls response creativity (0.0 to 1.0).	Keep it low (e.g., 0.2) for factual support assistants.
Max Tokens	Limits how long the AI response can be.	Prevents unnecessarily long and expensive completions.

Agent Templates

If you need a quick starting point, you can install pre-configured Agent Templates. These provide optimized system prompts and model settings for common roles (like Customer Support or Lead Triage) that you can install and customize for your workspace.

Tracking Scenario Usage

Before editing or deleting a preset, check the Usage Badge on the right side of the list:

• No usage: Safe to delete or modify.
• N scenarios: Currently used by active scenarios. Click the badge to view the exact list of published scenarios using this preset in the usage drawer.