---
title: "Start Alert AI Assistant thread"
method: POST
path: "/api/v2/AiAssistants/start"
tags: ["AlertAiAssistant"]
---

# Start Alert AI Assistant thread

`POST /api/v2/AiAssistants/start`

Generate the first alert-focused response for a request scoped to a specific alert. This compatibility path remains `/api/v2/AiAssistants/start`, but the request payload is alert-specific and requires `providerConfig` plus alert context in `conversationContext`. Provide either `conversationContext.metaInstructions` with an alert reference such as `{alertId:12345}`, or provide one or more `conversationContext.entityPayloads`. Requests that provide neither are rejected.

## Headers

- `Authorization` string, required
- `Content-Type` string, required
- `Org-Access` boolean
- `Account-Name` string

## Request body

- AiAssistantsStartSchema
  - `assistantType` string — Optional compatibility field for the existing public route. If supplied, the only supported value is `general`.
  - `conversationContext` object, required — Alert-specific context for the assistant request. Provide either `metaInstructions` with an alert reference such as `{alertId:12345}` or at least one entry in `entityPayloads`. If neither is provided, the request is rejected.
    - `metaInstructions` string — Alert reference used to scope the assistant to the target alert. Use this when you want to pass the alert ID through instructions, for example `User is asking about this alert? {alertId:12345}`.
    - `entityPayloads` object[] — Alert-specific payload objects supplied to the assistant. Use this when you want to send the alert payload directly instead of relying on `metaInstructions`.
    - `entities` object[] — Legacy entity list. Send an empty array when unused.
  - `providerConfig` object, required — Provider and model configuration for the assistant request.
    - `provider` string, required — LLM provider, such as `openai` or `bedrock`.
    - `modelConfig` object, required
      - `type` string, required — Model identifier, such as `default` (Bedrock — resolves to the current production model) or a pinned variant.

## Response `201`

No Error (AI Assistants start thread)

- object
  - `data` AiAssistantsStartResponseSchema
    - `data` unknown

## Other responses

- `4XX` — Client Error
- `5XX` — Internal Server Error

---

[API](https://skmtc.net/lacework/apis/forticnapp-api-2-0-documentation.md) · [All operations](https://skmtc.net/lacework/apis/forticnapp-api-2-0-documentation/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/lacework/forticnapp-api-2-0-documentation/versions/7015f76895f2/schema)