DocsOCR
Digitization
OCR
Data Digitization extracts structured data from documents using AI. Register a custom extraction model, run documents through it as a job, and review the results before they're saved.
Quick Start
- 1
Register a model
Connect a custom Azure Form Recognizer model — give it a Display Name, the Azure Model ID, and your Azure Endpoint.
- 2
Create a job
Upload one or more documents for batch processing against a registered model.
- 3
Let it process
The job extracts structured data from each document and tracks progress until every file is processed.
- 4
Review & approve
Once a job finishes, its output moves to Review, where extracted data is verified before it's saved.
Registering a model
Models are connected from Azure Form Recognizer, not trained inside NNIPA. You'll need a model already set up in Azure before you can register it here.
Display NameA human-readable name for the model, e.g. "IRS Form 1040".DescriptionOptional — describe what documents this model extracts.Azure Model IDThe model ID from Azure Form Recognizer, e.g. form1040-prod-v2.VersionOptional version label for the model, e.g. 1.0.Azure EndpointYour Azure Form Recognizer endpoint URL.Good to know
- A job can process a single document or a batch — the Files count on the Jobs page shows how many have completed out of the total.
- Jobs move through processing automatically; once done, they land in Review for verification.
- You can register multiple models for different document types and pick the right one per job.