> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# start • AI Agent SDK

> start: Start the agent interactively with verbose output.

# start

<div className="flex items-center gap-2">
  <Badge color="purple">Method</Badge>
</div>

> This is a method of the [**Agent**](../classes/Agent) class in the [**agent**](../modules/agent) module.

Start the agent interactively with verbose output.

Beginner-friendly execution. Defaults to verbose output with streaming
when running in a TTY. Use this for interactive/terminal usage where
you want to see output in real-time with rich formatting.

## Signature

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
def start(prompt: str) -> Any
```

## Parameters

<ParamField query="prompt" type="str" required={false}>
  The input prompt to process. If not provided, uses the  agent's instructions as the task (useful when instructions already describe what the agent should do). \*\*kwargs: Additional arguments: - stream (bool | None): Override streaming. None = auto-detect TTY - output (str): Output preset override (e.g., "silent", "verbose")
</ParamField>

### Returns

<ResponseField name="Returns" type="Any">
  Generator yielding response chunks

  * If not streaming: The complete response as a string
</ResponseField>

## Usage

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
# Minimal usage - instructions IS the task
    agent = Agent(instructions="Research AI trends and summarize")
    result = agent.start()  # Uses instructions as task

    # With explicit prompt (overrides/adds to instructions)
    agent = Agent(instructions="You are a helpful assistant")
    result = agent.start("What is 2+2?")  # Uses prompt as task
```

## Uses

* `substitute_variables`
* `isatty`
* `time_module.time`
* `chat`
* `threading.Thread`
* `chat_thread.start`
* `Panel`
* `Panel.fit`
* `Markdown`
* `Text`

## Notes

Unlike .run() which is always silent (production use), .start()
enables verbose output by default when in a TTY for beginner-friendly
interactive use. Use .run() for programmatic/scripted usage.

## Used By

* [`Agent.start`](../functions/Agent-start)
* [`AgentTeam.run`](../functions/AgentTeam-run)
* [`AutoAgents.start`](../functions/AutoAgents-start)

## Source

<Card title="View on GitHub" icon="github" href="https://github.com/MervinPraison/PraisonAI/blob/main/src/praisonai-agents/praisonaiagents/agent/agent.py#L6112">
  `praisonaiagents/agent/agent.py` at line 6112
</Card>