Simulate conversation

1 from elevenlabs import ( 2 AgentConfigDbModel, 3 ConversationSimulationSpecification, 4 ElevenLabs, 5 ) 6 7 client = ElevenLabs( 8 api_key="YOUR_API_KEY", 9 ) 10 client.conversational_ai.agents.simulate_conversation( 11 agent_id="21m00Tcm4TlvDq8ikWAM", 12 simulation_specification=ConversationSimulationSpecification( 13 simulated_user_config=AgentConfigDbModel( 14 first_message="Hello, how can I help you today?", 15 language="en", 16 ), 17 ), 18 )

1 { 2 "simulated_conversation": [ 3 { 4 "role": "user", 5 "time_in_call_secs": 1, 6 "message": "message", 7 "tool_calls": [ 8 { 9 "request_id": "request_id", 10 "tool_name": "tool_name", 11 "params_as_json": "params_as_json", 12 "tool_has_been_called": true 13 } 14 ], 15 "tool_results": [ 16 { 17 "request_id": "request_id", 18 "tool_name": "tool_name", 19 "result_value": "result_value", 20 "is_error": true, 21 "tool_has_been_called": true 22 } 23 ], 24 "feedback": { 25 "score": "like", 26 "time_in_call_secs": 1 27 }, 28 "llm_override": "llm_override", 29 "rag_retrieval_info": { 30 "chunks": [ 31 { 32 "document_id": "document_id", 33 "chunk_id": "chunk_id", 34 "vector_distance": 1.1 35 } 36 ], 37 "embedding_model": "e5_mistral_7b_instruct", 38 "retrieval_query": "retrieval_query", 39 "rag_latency_secs": 1.1 40 } 41 } 42 ], 43 "analysis": { 44 "call_successful": "success", 45 "transcript_summary": "transcript_summary", 46 "evaluation_criteria_results": { 47 "key": { 48 "criteria_id": "criteria_id", 49 "result": "success", 50 "rationale": "rationale" 51 } 52 }, 53 "data_collection_results": { 54 "key": { 55 "data_collection_id": "data_collection_id", 56 "rationale": "rationale" 57 } 58 } 59 } 60 }

Run a conversation between the agent and a simulated user.

Path parameters

agent_idstringRequired

The id of an agent. This is returned on agent creation.

Request

This endpoint expects an object.

simulation_specificationobjectRequired

A specification detailing how the conversation should be simulated

extra_evaluation_criterialist of objectsOptional

A list of evaluation criteria to test

Response

Successful Response

simulated_conversationlist of objects

analysisobject

Path parameters

Headers

Request

Response

Errors