Skip to main content
Properly configuring your agent is essential for accuracy, performance, and reliability. The Settings section lets you manage prompts, models, query behavior, and search precision.

Settings Navigation


Agent Settings: Part 1

Configure the base models for your agent.
  • Navigate to Agent Settings
  • Update the Agent name (if needed)
  • Choose your Chat model
  • Select an Embedding model
  • Save the settings
Agent Settings Part 1

Agent Settings: General Configuration


Agent Settings: Part 2 – Prompts & Library

βœ… Configure Prompt Behavior

  • Prompt for Own Data: Agent only responds using provided data. No hallucinations β€” replies with β€œI don’t know” if unsure.
  • Prompt for Web Data: Good for standalone, follow-up questions without context.
  • Prompt Library: Use system-defined prompts for use cases like Interviewer, Educator, or Assistant.
Prompt Configuration

Prompt Configuration Panel in Agent Settings

Let’s now look at how to explore and apply these predefined prompts.

Using the Prompt Library

1

Browse and Apply Predefined Prompts

Click Prompt lib to access the built-in library of system prompts.
Select a category that matches your agent’s purpose β€” e.g., Position Interviewer, Storyteller, FAQ Agent.
Prompt Library View

Prompt Library: Browse Built-in Prompt Options

2

Preview and Confirm Selection

After selecting a prompt, preview its description and sample usage.
Click Use Prompt to confirm and apply it to your agent.
Prompt Confirmation Dialog

Prompt Confirmation: View & Confirm Prebuilt Prompt


Agent Settings: Part 3 - Temperature

Control the creativity or precision of your agent responses:
  • 0.0 – 0.3: Very accurate and focused
  • 0.4 – 0.7: Balanced response generation
  • 0.8 – 1.0: More imaginative and open-ended
Temperature Setting

Agent Settings: Adjusting Response Temperature


Starter Questions

Let users quickly start conversations with predefined questions.

How to Add Starter Questions

  • Go to Agent Settings β†’ Starter Questions
  • Add up to 4 questions relevant to your agent
  • Click Add and then Save
Starter Config

Starter Questions: Configuration Screen

Add up to 4 questions relevant to your agent:
Starter Config 2

Starter Questions: Configure Screen

To preview how your saved starter questions look in your console and user experience:
Playground View with Starter Questions

Starter Questions: Display in Playground


Query Parameters

Query parameters allow you to inject dynamic data β€” like user ID, department, or context β€” directly into your agent’s logic.

How to Add Query Parameters

  • Go to Agent Settings
  • Click Add Query Parameter
  • Provide a name and reference in your prompt
  • Save the changes
Add Query Param

Query Parameters: Configuration in Settings

Then, test the parameter live in the Playground UI:
Param Visible to User

Query Parameters: Visible in Playground


🎯 Semantic Search Similarity

Control how precisely the agent matches your knowledge base.

Relevance Thresholds

  • β‰₯ 0.2: Broad, flexible match
  • β‰₯ 0.5: Moderate contextual relevance
  • β‰₯ 0.7: Strict precision
Change it under Agent Settings with a simple dropdown.
Semantic Score Slider

Semantic Scoring: Changing the Match Threshold


πŸ“© Lead Collection

You can now enable lead collection directly from the Settings panel.

How it Works

  • When Enable Lead Collection is turned on, a contact form will be shown inside the agent widget.
  • This form gathers user information like name, email, and optional notes.
  • It helps convert casual visitors into potential leads or customers.
Lead Collection Toggle

Settings: Lead Collection Toggle

Lead collection is a powerful way to capture qualified interest and convert chat interactions into actionable sales or support opportunities.
Configure these settings thoughtfully to ensure your agent delivers helpful, relevant, and contextually aware responses every time.
⌘I