> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs-parschat.parstechai.com/api-reference/webhook-events/message-generating-webhook/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs-parschat.parstechai.com/_mcp/server.

# message.generating

POST 

The AI has started composing a reply. Show a typing indicator.

Hide it when the matching `message` arrives. **There is no explicit stopped event**: a generation that fails produces no message, so treat it as a timeout after a few seconds.

This is the AI. A human typing is the separate `is_typing` event.

Reference: https://docs-parschat.parstechai.com/api-reference/webhook-events/message-generating-webhook

## Request

### Payload

- `schema_version` (string, optional) — Semver for the payload shape. Fields are added without a major bump; nothing is removed or repurposed without one.
- `event` (string, optional) — Which event this is.
- `delivery_id` (string, optional) — Unique per delivery attempt. Log it, and use it to detect a redelivery.
- `occurred_at` (datetime, optional) — When the event happened.
- `robot_id` (string, optional) — The robot it belongs to.
- `external_id` (string, optional) — Your own id for that robot, echoed on every event so you never need a lookup table.
- `channel_id` (string, optional) — The channel it arrived on.
- `data` (WebhooksMessageGeneratingWebhookPayloadContentApplicationJsonSchemaData, optional)

## Types

### WebhooksMessageGeneratingWebhookPayloadContentApplicationJsonSchemaData

- `conversation_id` (string, optional) — The conversation the AI is answering.