- Large or complex documents
- Batch processing
- Workloads where extraction time may exceed synchronous limits
- An extraction job is created
- A GUID is returned immediately
- Processing continues in the background
- Results are retrieved by polling the job status endpoint