simulate

Generate AI predictions and responses
RUN

simulate

Generates predictions of how a person would think, feel, communicate, and act in any scenario you describe. This is where training data becomes actionable insights.

Using the trained snapshot, run AI simulations to predict responses to any scenario. Each simulation completes in seconds, and you can run unlimited simulations once a snapshot is complete.

Parameters

mindId

Typestring
RequiredYes
DescriptionThe mind ID from the create_mind response

scenario.message

Typestring
RequiredYes
DescriptionThe scenario or question to simulate. Be specific and include emotional context, time constraints, and desired outcomes for best results.

Request

1{
2 "mindId": "abc-123-def-456",
3 "scenario": {
4 "message": "A customer is frustrated because their order is delayed by two weeks. They need the product for an event tomorrow. The item is currently out of stock. How would you respond?"
5 }
6}

Response

1{
2 "message": "Simulation completed successfully"
3}

Response Properties

PropertyTypeDescription
messagestringStatus message confirming simulation completion
Response Format

The simulation response contains the AI-generated prediction of how the person would respond. The exact format may vary based on the model used, but always includes the predicted response text.

Writing Effective Scenarios

The quality of your simulation depends heavily on scenario specificity.

Vague scenario:

"How would you handle a complaint?"

Too general—lacks context, constraints, and emotional state.

Specific scenario:

"A customer received a damaged product 3 days before their wedding.
They need a replacement by tomorrow, but the item is out of stock.
They're understandably upset and considering canceling the order.
How would you respond?"

Includes context, time constraints, emotional state, and desired outcome.

Include these elements:

  • Emotional context: frustrated, confused, urgent, pleased
  • Time constraints: tomorrow, by Friday, immediately
  • Specific details: order numbers, product names, amounts
  • Constraints: out of stock, over budget, policy limits
  • Desired outcome: resolution, information, escalation

Common Use Cases

Train representatives on objection handling, discovery calls, and closing techniques.

Example:

"A prospect says our product is 3x more expensive than
the competitor they're currently using. They like our
features but can't justify the cost difference to their
CFO. How would you position the value?"

What you get:

  • Authentic objection-handling techniques
  • Value proposition phrasing
  • Discovery questions to uncover needs
  • Closing strategies

Quality assurance testing for customer complaints, technical issues, and policy questions.

Example:

"A customer's subscription was charged twice this month.
They're threatening to leave and post negative reviews.
They've been a loyal customer for 3 years. How would
you resolve this?"

What you get:

  • Empathy and de-escalation techniques
  • Problem-solving approach
  • Compensation strategies
  • Retention tactics

Prepare for presentations by anticipating questions, concerns, and objections.

Example:

"I'm proposing a $500K investment in new infrastructure
that will take 18 months to show ROI. Based on past
board meetings, what questions and concerns should I
expect?"

What you get:

  • Likely questions and concerns
  • Communication style preferences
  • Key metrics to emphasize
  • Pre-emptive answers to prepare

Error Responses

1{
2 "error": "Not Found",
3 "message": "Mind with ID 'abc-123' does not exist"
4}

Solution: Verify the mindId is correct and ensure you have a completed snapshot for this mind.

1{
2 "error": "Bad Request",
3 "message": "No completed snapshot found for this mind"
4}

Solution: Verify snapshot training is complete using get_snapshot_status. Wait for snapshot to finish processing.

1{
2 "error": "Bad Request",
3 "message": "Scenario message is required"
4}

Solution: Ensure the scenario.message parameter is provided with a non-empty string.

Best Practices

Start Specific
  • Include emotional context and constraints
  • Provide relevant background information
  • Specify desired outcomes
  • Add time pressure when relevant
  • Run multiple variations of scenarios
  • Look for patterns in responses
  • Test edge cases and unusual situations
  • Compare responses across scenarios
  • Start with known scenarios to validate accuracy
  • Gradually introduce novel situations
  • Document unexpected responses
  • Verify critical predictions manually
  • Use reasoning to understand predictions

See Also