​
What is LAO?

Lyzr Agent Orchestration (LAO) is a framework to define and execute workflows using Lyzr Agents. Workflows are defined in JSON format and executed through the LAO engine.

​
JSON Format Structure

{
  "tasks": [],
  "default_inputs": {},
  "flow_name": "",
  "run_name": ""
}

​
Field Descriptions:

  • tasks: An array of task objects.
  • default_inputs: Input values reused across the workflow.
  • flow_name: Name assigned to the workflow.
  • run_name: Identifier for the particular execution of the workflow.

​
Defining a Task

Each task is an object with the following structure: 
{
  "name": "",
  "tag": "",
  "function": "",
  "params": {}
}

​
Task Fields:

  • name: A unique identifier for the task.
  • tag: (Optional) A readable label for better clarity.
  • function: The action to execute. (e.g., call_lyzr_agent, api_call)
  • params: The parameters required for the function to execute.

​
Defining Inputs in params

You can define parameter values in three different ways:

​
1. Direct Input

"param_name": "value"

​
2. Input from default_inputs

"param_name": {
  "input": "input_key"
}

​
3. Dependency on Another Task

"param_name": {
  "depends": "task_name"
}

​
Example: Calling a Lyzr Agent

Use the call_lyzr_agent function to call a Lyzr Agent.

​
Step 1: Define Agent Config in default_inputs

"mail_generator_agent_config": {
  "user_id": "",
  "session_id": "",
  "agent_id": "",
  "api_key": "",
  "api_url": "https://agent.api.lyzr.app/v2/chat/",
  "agent_name": "mail_generator_agent"
}
💡 To get agent_id, session_id, and user_id, please refer to the API documentation or check under the Inference tab located in the Agent API section of the Agent Dashboard on Lyzr Studio.

​
Step 2: Task Definition

{
  "name": "task_1",
  "tag": "Mail Generator Agent",
  "function": "call_lyzr_agent",
  "params": {
    "config": {
      "input": "mail_generator_agent_config"
    },
    "user_message": "Create an email on the topic AI.",
    "email_tone": "Formal"
  }
}

​
Step 3: Full Workflow JSON

{
  "tasks": [
    {
      "name": "task_1",
      "tag": "Mail Generator Agent",
      "function": "call_lyzr_agent",
      "params": {
        "config": {
          "input": "mail_generator_agent_config"
        },
        "user_message": "Create an email on the topic AI.",
        "email_tone": "Formal"
      }
    }
  ],
  "default_inputs": {
    "mail_generator_agent_config": {
      "user_id": "",
      "session_id": "",
      "agent_id": "",
      "api_key": "",
      "api_url": "https://agent.api.lyzr.app/v2/chat/",
      "agent_name": "mail_generator_agent"
    }
  },
  "flow_name": "email_generator_flow",
  "run_name": "run_1"
}

​
Example: Calling an External API

Use the api_call function to make API calls.

​
Step 1: Define API Config in default_inputs

"weather_api_config": {
  "url": "https://api.openweathermap.org/data/2.5/weather",
  "method": "GET",
  "headers": {
    "accept": "application/json"
  }
}

​
Step 2: Task Definition

{
  "name": "task_1",
  "tag": "Get Current Weather",
  "function": "api_call",
  "params": {
    "config": {
      "input": "weather_api_config"
    },
    "QUERY_q": {
      "input": "city"
    },
    "QUERY_units": "metric"
  }
}

​
Step 3: Full Workflow JSON

{
  "tasks": [
    {
      "name": "task_1",
      "tag": "Get Current Weather",
      "function": "api_call",
      "params": {
        "config": {
          "input": "weather_api_config"
        },
        "QUERY_q": {
          "input": "city"
        },
        "QUERY_units": "metric"
      }
    }
  ],
  "default_inputs": {
    "weather_api_config": {
      "url": "https://api.openweathermap.org/data/2.5/weather",
      "method": "GET",
      "headers": {
        "accept": "application/json"
      }
    }
  },
  "flow_name": "weather_flow",
  "run_name": "run_1"
}

​
Summary

With Lyzr Agent Orchestration, you can build powerful and intelligent workflows using Lyzr Agents and API integrations. Chain tasks together, reuse inputs efficiently, and automate complex logic in a clean and modular way.