Skip to main content

​
What is Copilot?

Copilot is an AI-powered prompt editor integrated into the Advanced Editor. Instead of just giving you suggestions, Copilot directly edits and improves your agent’s system prompt based on your instructions. It analyzes your current prompt, generates improvements, and applies the changes automatically.
Copilot doesn’t just suggest - it edits. Describe what you want to improve and watch as it modifies your system prompt in real-time.

​
Accessing Copilot

Copilot is located in the right panel of the Advanced Editor. Click the “Copilot” tab or press ⌘J to access it quickly.

​
Welcome Screen

When you first open Copilot, you’ll see a clean welcome interface with quick action buttons for common tasks:
1

Analyze & Improve

Click to have Copilot review your entire prompt and suggest comprehensive improvements.Analyzes structure, clarity, instructions, and effectiveness automatically.
2

Refine Tone

Makes your prompt more concise and professional.Removes redundancy and improves conversational style.
3

Add Output Fields

Automatically adds output parameters for lead qualification.Includes fields like name, email, phone, and interest level.
4

Create Evaluations

Sets up performance measurements for your agent.Creates metrics for conversation quality, goal completion, and customer satisfaction.

​
How It Works

Copilot operates in three phases when you request a change:
PhaseWhat Happens
🔍 AnalyzingCopilot reads your current system prompt and understands the context
⚙️ GeneratingIt creates the improved version based on your request
✏️ EditingChanges are applied directly to your system prompt
After completion, you’ll see a summary of all changes made, which you can accept or reject.

​
AI Suggestions for Input Parameters & Output Settings

Copilot now provides intelligent suggestions for input parameters, output settings, and evaluations that appear directly in their respective configuration panels. These suggestions are contextual to your current agent setup and conversation needs.
When Copilot generates suggestions for parameters or evaluations, they appear as pending changes in the relevant panels with purple highlighting and accept/reject buttons.

​
Input Parameter Suggestions

When Copilot analyzes your agent, it can automatically suggest new input parameters based on:
  • Agent Role: Appointment schedulers need calendar preferences, sales agents need budget information
  • Conversation Context: Referenced calls that show missing information
  • Industry Standards: Common parameters for your agent’s domain
1

Generate Parameter Suggestions

Ask Copilot to “Add input parameters for appointment scheduling” or use a quick action button.
2

Review Pending Changes

Navigate to the Input Parameters panel to see suggested parameters with purple highlighting.
3

Accept or Reject

Each suggestion shows operation type (Add/Modify/Remove), parameter details, and accept/reject buttons.
4

Bulk Actions

Use “Accept All” to apply all suggestions or reject individual ones selectively.

​
Output Settings Suggestions

Copilot suggests output parameters to capture key conversation data: 
@call:abc123 The agent collected the customer's name and interest level 
but we're missing their preferred contact method. Add output fields to capture this.
Copilot will analyze the call context and suggest appropriate output parameters like:
  • Contact preferences (email, phone, text)
  • Lead qualification scores
  • Next step requirements
  • Custom fields based on conversation patterns

​
Evaluation Suggestions

Performance measurements are automatically suggested based on your agent’s goals:
  • Conversation Quality: Measures clarity, professionalism, and engagement
  • Goal Achievement: Tracks completion of primary objectives
  • Customer Satisfaction: Evaluates customer experience and satisfaction
  • Process Adherence: Ensures agent follows required procedures

​
Visual Indicators for Pending Changes

​
Purple Highlight System

When Copilot generates suggestions for input parameters, output settings, or evaluations, you’ll see distinctive visual indicators:
Purple-highlighted banners appear at the top of relevant panels showing:
  • Copilot sparkle icon with “Copilot Suggestions” title
  • Summary of pending changes (e.g., “3 new parameters, 1 evaluation”)
  • Accept All/Reject buttons for bulk actions

​
Change Types and Colors

Change TypeBadge ColorDescription
Add🟢 GreenNew parameters, evaluations, or settings
Modify🟡 YellowChanges to existing configurations
Remove🔴 RedDeletion of outdated or redundant items

​
Accepting and Rejecting Suggestions

​
Individual Change Management

1

Review Each Suggestion

Each change card shows the operation type, parameter details, and impact on your agent.
2

Accept Individual Changes

Click the “Accept” button on specific suggestions you want to keep.
3

Reject Unwanted Changes

Click “Reject” to dismiss suggestions that don’t fit your needs.
4

See Live Updates

Accepted changes immediately appear in your configuration, rejected ones are removed from pending.

​
Bulk Change Management

Use the banner controls for efficient bulk operations: 
Accept All - Apply all pending suggestions at once
Reject All - Dismiss all pending suggestions
Bulk actions cannot be undone. Review individual suggestions before using “Accept All” to ensure they align with your agent’s requirements.

​
Change Persistence

  • Pending changes persist across editor sessions until accepted or rejected
  • Auto-save functionality preserves your accept/reject decisions
  • Version history tracks all applied changes for rollback if needed

​
Voice Recording

Copilot supports voice input for hands-free prompt editing:

​
Using Voice Commands

1

Start Recording

Hold the microphone button in the input field, or double-press Shift to begin voice recording.
2

Speak Your Request

Clearly describe what you want Copilot to improve or change in your prompt.
3

Stop Recording

Release the mic button or press Shift again to end recording.
4

Review & Send

The transcribed text appears in the input field. Review and send your request.

​
Voice Input Tips

  • Speak clearly and at a normal pace
  • Use natural language - describe changes as you would to a colleague
  • Voice input is transcribed to text before processing
Voice recording works best in quiet environments. If transcription is unclear, you can edit the text before sending.

​
Call Context Integration

Reference specific calls to inform prompt improvements using the @call:ID syntax:

​
Using Call References

@call:abc123 The agent failed to handle the budget question. 
Add instructions to better address pricing concerns.
When you reference a call, Copilot loads the transcript and uses it to understand what went wrong and how to improve the prompt.

​
Call Context Indicators

Call references appear with special indicators in the chat:
  • @call:abc123 - Shows as a clickable reference with call metadata
  • Hover to see call details (duration, outcome, timestamp)
  • Click to open the full call transcript
1

Find a Problematic Call

Identify a call where the agent didn’t perform well in your Calls dashboard.
2

Copy the Call ID

Use the call ID from the URL or copy from the call details.
3

Reference in Copilot

Type @call:CALL_ID followed by your improvement request.
4

Review Context-Aware Changes

Copilot edits the prompt based on the specific call context.

​
Describing Changes

In the input field, describe what you want to change. Be specific about your goals.

​
Effective Requests

Add instructions for handling pricing objections

Make the greeting more friendly and casual

Add a section about appointment confirmation procedures

@call:xyz789 Improve how the agent handles "I need to think about it" responses

​
What Copilot Edits

Copilot can modify any part of your system prompt:
ElementExamples
InstructionsAdd new behaviors, modify existing rules
ToneAdjust formality, friendliness, professionalism
StructureReorganize sections, add headers
GuardrailsAdd safety limits, scope boundaries
ScenariosAdd handling for specific situations
LanguageImprove clarity, remove ambiguity
Output FieldsAdd data collection parameters
EvaluationsCreate performance measurement criteria

​
Next Step Suggestions

After Copilot completes a change, you’ll see suggested next steps to further improve your agent:

​
Types of Suggestions

​
Using Suggestions

Click any suggestion to automatically populate the input field with that request. You can:
  • Use suggestions as-is
  • Modify them before sending
  • Combine multiple suggestions
  • Dismiss suggestions you don’t need

​
Analyzing Tickets

For bulk analysis, use the Tickets panel:
1

Open Tickets Panel

Click the + button and select “Analyze tickets”.
2

Select Tickets

Choose multiple calls/tickets to analyze.
3

Run Analysis

Copilot identifies patterns across selected tickets.
4

Apply Improvements

Suggested prompt changes based on common issues.

​
Reviewing Changes

After Copilot edits your prompt:

​
Change Summary

You’ll see a completion summary showing:
  • Title of the changes made
  • Duration of the operation
  • List of specific modifications
  • Suggested next steps

​
Accept or Reject

Changes appear in the Monaco editor with highlighting:
  • Accept: Keep the changes in your prompt
  • Reject: Revert to the previous version
Always review changes before accepting. Test in the Playground to verify the improvements work as expected.

​
Conversation History

Continue refining with follow-up requests: 
User: Add instructions for handling price objections

Copilot: [Makes changes]

User: Good, but make it less aggressive

Copilot: [Refines the changes]

User: Perfect, now add a fallback for if they still object

Copilot: [Adds additional handling]
Start a new conversation using the menu option to reset context when switching to a different topic.

​
Voice Input Best Practices

​
Recording Environment

  • Use a quiet space with minimal background noise
  • Position yourself close to your device’s microphone
  • Speak at a normal, conversational pace

​
Effective Voice Commands

"Add greeting instructions for appointment scheduling calls"

"Reference call A-B-C-1-2-3. The customer was confused about pricing. 
Improve the pricing explanation section."

"Make the entire prompt sound more friendly and conversational"

​
Voice vs. Text Input

Use Voice WhenUse Text When
Making quick improvementsWriting complex, technical instructions
Referencing call transcriptsAdding specific code or formatting
Describing tone changesMaking precise edits to existing content
Brainstorming improvementsIncluding URLs or technical specifications

​
Best Practices

1

Be Specific About Changes

Instead of “make it better”, describe exactly what to improve:Good: “Add a section for handling refund requests with a 3-step process”Vague: “Improve customer service handling”
2

Use Real Call Context

Reference actual failed calls when improving specific scenarios. This gives Copilot concrete examples of what went wrong.Use @call:ID syntax to provide context from real conversations.
3

Review Parameter Suggestions

Check the Input Parameters and Output Settings panels after requesting parameter changes.Purple highlights indicate pending suggestions that need your approval.
4

Accept Changes Strategically

Don’t blindly accept all suggestions. Review each parameter or evaluation to ensure it fits your specific use case.
5

Test Comprehensive Changes

When accepting multiple parameter or evaluation suggestions, test in Playground to verify everything works together.
6

Use Voice for Natural Requests

Voice input works especially well for describing parameter needs: “The agent needs to collect the customer’s budget range and timeline preferences.”
7

Follow Visual Cues

Purple notification indicators guide you to panels with pending changes. Don’t miss suggestions in other panels.
8

Iterate in Small Steps

Make incremental changes rather than trying to fix everything at once. This makes it easier to identify what works.

​
Common Editing Tasks

​
Adding New Behaviors

Add instructions for the agent to always confirm 
the customer's phone number before ending the call

​
Improving Specific Scenarios

@call:abc123 Improve how the agent handles when 
customers ask to speak with a manager

​
Adjusting Communication Style

Make the entire prompt more conversational and less formal. 
The agent should sound like a helpful friend, not a robot.

​
Adding Parameters and Output Fields

Add input parameters to capture customer preferences and 
output fields to track lead qualification scores

​
Creating Comprehensive Evaluations

Create evaluations to measure: conversation quality, 
goal achievement, customer satisfaction, and proper procedure following

​
Adding Guardrails

Add clear boundaries: the agent should never discuss 
competitor products or make promises about delivery dates

​
Context-Based Parameter Updates

@call:xyz789 The customer mentioned their budget but we didn't capture it. 
Add input and output parameters for budget tracking.

​
Keyboard Shortcuts

ShortcutAction
⌘JOpen/focus Copilot panel
Shift + ShiftStart/stop voice recording
⌘EnterSend message to Copilot
EscapeClose Copilot suggestions
⌘3Navigate to Input Parameters panel
⌘7Navigate to Output Settings panel
⌘6Navigate to Evaluations panel

​
Next Steps

Input Parameters

Configure input parameters with AI suggestions

Output Settings

Set up output fields and webhooks

Evaluations

Create performance measurements

Playground

Test your improved prompts