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

# Start step

> Set up the step every call begins with: an opening message in Conversation mode, or an API call first in Tool mode, plus its interruption settings.

The Start step is where every call to a flow agent begins. It either greets the caller and listens (**Conversation mode**) or calls one of your APIs before anyone speaks (**Tool mode**). Every flow has exactly one Start step; you can't delete or copy it, and it isn't in the **Add node** list. On the canvas it's the one step in lime, with a solid lime tab and frame, so you can always find where calls begin.

![The Greeting Start step on the canvas with its lime tab reading Start (conversation) and lime frame, next to the Verify the order Conversation step with a light blue tab](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/flow/steps/start-step-card-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=11aeb1d61468d1136e082db171b43732)

## Choose a mode

The first field in the panel is **Start node type** ("Choose whether this start node should behave as a conversation or tool node").

| Option | When to use it |
| - | - |
| **Conversation mode** (default) | "Start with a conversation message to the user". Use it for most agents: the agent greets the caller and the flow goes on from what they say. |
| **Tool mode** | "Start by executing a tool or API call". Use it when you need data before the first word, for example to look up the caller by their number. |

The panel title follows the mode: **Start (conversation)** or **Start (tool)**.

![The Start step panel in Conversation mode: Start node type, the opening Message as a Prompt, and the Transitions, Interruption, Variables and Global sections](https://mintcdn.com/hamsa-452fa79a/Rl78D_qJIpinmg19/images/agents/flow/steps/start-step-light.webp?fit=max\&auto=format\&n=Rl78D_qJIpinmg19\&q=85\&s=b2437b772bf6896c7c20651a1901ebd9)

Switching modes after you've set one up opens **Change node type**, which warns that the switch clears the current mode's configuration, including its transitions. Click **Continue anyway** to switch.

## Conversation mode

### Fields

| Field | What it does | Default |
| - | - | - |
| **Message** | What the agent says first. Choose **Static** to speak the text as written, or **Prompt** to give the agent instructions and let it write the greeting. Use `{{variable_name}}` to include variables. Required: "Message is required". | **Prompt** |
| **Interrupt the opening message** | "Let the user interrupt the agent while it plays the opening message." | See below |
| **Interruption** | **Inherit**, **Enable** or **Disable**. On the Start step, only **Disable** has its own effect: it stops the caller from interrupting, whatever the toggle says. **Inherit** and **Enable** follow **Interrupt the opening message**. | **Inherit** |

When **Interruption** is **Disable**, the toggle greys out with a note that it has no effect.

A new flow doesn't set **Interrupt the opening message**, and on a call an unset toggle lets the caller interrupt. To make sure the opening can't be interrupted, set **Interruption** to **Disable**.

### Transitions

A Start step in Conversation mode starts with no transitions. You can add:

| Type | Moves on when |
| - | - |
| **Prompt** | The conversation matches a condition you describe in words. |
| **When a live variable matches** | A live variable meets your rules. |
| **DTMF input** | The caller presses a key. |
| **After user replies** | The caller finishes their first reply, whatever they say. |

**Skip if** and **Auto** aren't offered on the Start step. See [Transitions](https://docs.tryhamsa.com/agents/flow/transitions.md) for each type.

### Variables

In the **Variables** section you can add a **Live variable**, a **Variable on node exit**, a **Local variable** or **DTMF capture** ("Capture a sequence of DTMF digits and store it in a variable. Use \* or # on transitions as termination keys."). The agent's own live variables from **Global settings** are listed there too, each with a switch to turn it off for this step. See [Live variables](https://docs.tryhamsa.com/agents/variables/live-variables.md), [Extract values from the conversation](https://docs.tryhamsa.com/agents/variables/extracted-variables.md) and [Keypad input](https://docs.tryhamsa.com/agents/flow/keypad-input.md).

## Tool mode

### Fields

| Field | What it does | Default |
| - | - | - |
| **Processing message** | What the agent says while the API call runs. **Static** speaks your text; **Prompt** lets the agent write it. Leave it empty to say nothing. | **Static** |
| **Tool configuration** | "Configure the tool to execute when the workflow starts". Click **Select tool** to pick one of your API tools, then adjust its settings for this step. Required: "Please select a tool when using tool mode". | None |
| **Test tool** | Runs the tool with sample values so you can see the response and set up response variables. | |

Tool configuration works as on a [Tool step](https://docs.tryhamsa.com/agents/flow/steps/api-request.md#tool-configuration). Tool mode offers API tools only.

The panel shows **Manual configuration required**: at the start of a call there's no conversation yet, so the agent can't fill tool parameters from what the caller said. Each parameter shows a **Value** field ("Manual value required" on required ones): type a fixed value, or insert a variable with `{{variable_name}}`, for example a value you pass when the call starts (see [Pass values when a call starts](https://docs.tryhamsa.com/agents/variables/dynamic-variables.md)) or a system variable such as the caller's number (see [System variables](https://docs.tryhamsa.com/agents/variables/system-variables.md)).

### Transitions

Tool mode creates two fixed transitions, and you can't add others:

| Transition | Taken when |
| - | - |
| **On Success** | The API answers with a 2xx status. |
| **On Failure** | The API answers with any other status, can't be reached, or takes longer than 30 seconds. |

### Variables

Add a **Tool response variable** to save a value from the response, or a **Local variable**. See [Tool step](https://docs.tryhamsa.com/agents/flow/steps/api-request.md#save-values-from-the-response).

## How it works on a call

* The Start step runs as soon as the call connects.
* **Conversation mode:** the agent speaks the opening message (or writes one from your prompt), then waits for the caller. The step stays active, and the agent keeps talking with the caller, until one of its transitions fires.
* **Tool mode:** the step runs exactly like a [Tool step](https://docs.tryhamsa.com/agents/flow/steps/api-request.md#how-it-works-on-a-call), then follows **On Success** or **On Failure**. There is no **Put on hold** option on the Start step.

## Good to know

* The Start step has a **Global** section like other steps, but no **Confirm before proceeding**.
* A Start step in Conversation mode has no **Auto** transition, so it can't move on right after the greeting. To do work before the greeting, use Tool mode; to move on after the caller's first answer, add **After user replies**.

## Related

> [card] [Conversation step](https://docs.tryhamsa.com/agents/flow/steps/conversation.md)
>
> The same message and interruption settings, for every step after the start.

> [card] [Tool step](https://docs.tryhamsa.com/agents/flow/steps/api-request.md)
>
> How API calls, processing messages and response variables work.

> [card] [Transitions](https://docs.tryhamsa.com/agents/flow/transitions.md)
>
> Every transition type and when it fires.

> [card] [Write a prompt that works on calls](https://docs.tryhamsa.com/agents/build/write-a-prompt.md)
>
> How to write instructions the agent follows on a call.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.