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

# Update Agent Flow

> Update an agent's conversation mode and flow configuration

# Update Agent Flow

Update an agent's conversation mode and (when applicable) the structured flow configuration. Submitting an invalid flow returns a 422 with details — use [Validate Agent Flow](/api-reference/endpoints/validate-agent-flow) to dry-run validation before saving.

## Endpoint

```
PUT /api/v1/agents/{agent_id}/flow
```

## Path parameters

<ParamField path="agent_id" type="string" required>Agent UUID.</ParamField>

## Request headers

<ParamField header="Authorization" type="string" required>
  Bearer token. Format: `Bearer talq_your_environment_token_here`
</ParamField>

<ParamField header="Content-Type" type="string" required>
  Must be set to `application/json`
</ParamField>

## Request body

<ParamField body="conversation_mode" type="string" required>
  Conversation mode. Options:

  * `llm_driven` — LLM-guided conversation (knowledge + actions). Send `flow_config: null` or omit.
  * `flow_driven` — node-graph driven. `flow_config` is required.
</ParamField>

<ParamField body="flow_config" type="object">
  Flow definition. Required when `conversation_mode` is `flow_driven`.

  <Expandable title="flow_config object">
    <ParamField body="initial_node_id" type="string" required>
      ID of the entry node. Must reference an existing node in `nodes`.
    </ParamField>

    <ParamField body="meta" type="object">
      Optional metadata: `{ "name": "string", "version": "string" }`.
    </ParamField>

    <ParamField body="nodes" type="array" required>
      Array of node objects. Each node has at minimum: `id` (string, unique), `type` (`say`, `ask`, `decision`, `action`, `transfer`, `end`), and type-specific fields (`content`, `next`, `branches`, etc.).
    </ParamField>

    <ParamField body="global_functions" type="array">
      Functions invocable from any node. Common entries: `end_conversation`, `transfer_call`, `repeat_question`.
    </ParamField>
  </Expandable>
</ParamField>

## Examples

<RequestExample>
  ```bash theme={null}
  # Switch to llm_driven mode
  curl -X PUT "https://app.talkover.ai/api/v1/agents/agent-uuid/flow" \
    -H "Authorization: Bearer talq_your_environment_token_here" \
    -H "Content-Type: application/json" \
    -d '{ "conversation_mode": "llm_driven" }'
  ```

  ```bash theme={null}
  # Set a flow-driven configuration
  curl -X PUT "https://app.talkover.ai/api/v1/agents/agent-uuid/flow" \
    -H "Authorization: Bearer talq_your_environment_token_here" \
    -H "Content-Type: application/json" \
    -d '{
      "conversation_mode": "flow_driven",
      "flow_config": {
        "initial_node_id": "node-greet",
        "meta": { "name": "Intake v1", "version": "1.0.0" },
        "nodes": [
          { "id": "node-greet", "type": "say", "content": "Hi!", "next": "node-end" },
          { "id": "node-end", "type": "end" }
        ],
        "global_functions": [
          { "name": "end_conversation", "description": "Ends the call" }
        ]
      }
    }'
  ```
</RequestExample>

## Response

<ResponseExample>
  ```json theme={null}
  {
    "success": true,
    "data": {
      "conversation_mode": "flow_driven",
      "flow_config": { . }
    }
  }
  ```
</ResponseExample>

## Error responses

### 422 — Invalid Flow

<ResponseExample>
  ```json theme={null}
  {
    "success": false,
    "message": "Flow validation failed",
    "errors": [
      "initial_node_id 'node-greet' not found in nodes",
      "Node 'node-qualify' references unknown next node 'node-bookx'"
    ]
  }
  ```
</ResponseExample>

## Notes

* Updating the flow does not change the agent's status — but published agents need to be re-published if the flow change is breaking.
* See [List Flow Versions](/api-reference/endpoints/list-agent-flow-versions) to track historical configurations.