Introducing Parse Jobs API: Process 1000+ Page Documents

Today we're excited to announce the Parse Jobs API — our solution for processing massive documents asynchronously. Upload a 2,000-page mortgage packet or medical record bundle, and we'll process it in the background and notify you when complete.

Why We Built This

Our original Parse API works great for individual documents (invoices, forms, reports), but customers kept asking: "What about our 500-page mortgage packets?" or "Can we process an entire box of medical records at once?"

The synchronous API isn't ideal for these use cases:

• Long-running requests risk timeouts
• Client has to keep connection open for minutes/hours
• Retrying a failed 1000-page job means reprocessing everything

How It Works

Step 1: Create a Job

POST /v1/parse/jobs
Content-Type: application/json

{
  "files": [
    "https://your-bucket.s3.amazonaws.com/mortgage-packet.pdf"
  ],
  "mode": "parse",
  "webhook_url": "https://your-app.com/webhooks/retriv",
  "metadata": {
    "loan_id": "L-2024-00123",
    "customer": "John Doe"
  }
}

Returns a job_id immediately. Processing starts in the background.

Step 2: We Process in the Background

Your document is broken into chunks and processed in parallel across our GPU cluster. Depending on size and complexity, this can take anywhere from 30 seconds to 2 hours.

Step 3: Get Notified via Webhook

POST https://your-app.com/webhooks/retriv
Content-Type: application/json

{
  "event": "job.completed",
  "job_id": "job_abc123",
  "status": "completed",
  "result_url": "https://api.retriv.ai/v1/parse/jobs/job_abc123/result",
  "metadata": {
    "loan_id": "L-2024-00123",
    "customer": "John Doe"
  },
  "stats": {
    "total_pages": 847,
    "chunks_created": 423,
    "processing_time_seconds": 142
  }
}

Step 4: Download Results

GET /v1/parse/jobs/job_abc123/result

Returns the same structure as the sync Parse API:
{
  "chunks": [...],
  "metadata": {...},
  "pages": [...]
}

Use Cases

Pricing

Parse Jobs are 30% cheaper than the synchronous Parse API:

A 1,000-page document costs 7,000 credits (vs 10,000 with sync API).

What's Next

This is just v1 of Parse Jobs. On our roadmap:

• Priority queue: Pay more to jump the line for time-sensitive documents
• Progress callbacks: Get updates every N pages processed
• Scheduled jobs: Process documents at off-peak hours for lower cost
• Batch extraction: Apply the same schema to multiple documents in one job