AI Analysis

Generate video summaries, chapters, and content moderation using AI.

Supported Analysis Types

Type	Output	Use Case
summary	Text summary of the video	SEO descriptions, search indexing
chapters	Timestamped chapter markers	Video navigation, table of contents
moderation	Content safety analysis	NSFW detection, compliance

Trigger Analysis

bash

curl -X POST /api/ai/videos/1/analyse \
  -H "Authorization: Bearer TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"type": "summary"}'

Response

{
  "data": {
    "id": 1,
    "asset_id": 1,
    "analysis_type": "summary",
    "status": "processing",
    "model": "anthropic/claude-sonnet-4-6"
  }
}

Analysis runs asynchronously as a background job. Poll the results endpoint or use webhooks to be notified when complete.

Get Results

bash

curl /api/ai/videos/1/analysis \
  -H "Authorization: Bearer TOKEN"

Response (summary)

{
  "data": [
    {
      "id": 1,
      "analysis_type": "summary",
      "status": "ready",
      "result": "This video demonstrates how to set up a Node.js development environment...",
      "total_tokens": 1250,
      "model": "anthropic/claude-sonnet-4-6",
      "created_at": "2026-03-29T12:00:00Z"
    }
  ]
}

Video Q&A (Streaming)

Ask questions about a video with streaming AI responses:

bash

curl -N /api/ai/videos/1/qa \
  -H "Authorization: Bearer TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"question": "What are the main topics covered?"}'

Response is Server-Sent Events (SSE):

text

data: {"token": "The"}
data: {"token": " video"}
data: {"token": " covers"}
data: {"token": " three"}
data: {"token": " main"}
data: {"token": " topics"}
...
data: [DONE]

Configuration

AI features use the Vercel AI Gateway. Configure in your .env:

.env

AI_GATEWAY_URL=https://ai-gateway.vercel.sh/v1
AI_GATEWAY_API_KEY=vck_your_key_here
AI_GATEWAY_MODEL=anthropic/claude-sonnet-4-6

The AI Gateway supports any LLM provider (Anthropic, OpenAI, etc.). Change the model string to switch providers.

← Webhooks