AI-Powered Document Processing
Stop Retyping Purchase Orders. Let OCR Read Them for You.
AtomIQ uses Azure Document Intelligence to read purchase order and invoice documents automatically, extract the data, and route it straight into your ERP — no manual transcription, no re-typing line items.
How the Workflow Actually Runs
Real workflow architecture, pulled from an actual AtomIQ workflow configuration.
📄
Document Arrives
File location / blob
→
🔎
Azure Document Intelligence
Analyze Document (Invoice model)
→
{ }
Inline Transform
Clean and map fields
→
↳
JQ Transform
Structure for downstream use
→
➜
Your ERP
NetSuite / JD Edwards / SAP
1
Document lands in a watched location
A purchase order or invoice arrives via FTP, blob storage, or a direct document URL.
2
Azure Document Intelligence analyzes it
The configured document model (e.g. Invoice) extracts structured fields — line items, totals, dates, vendor details — from the source file.
3
Data is cleaned and transformed
Inline JS and JQ transformation steps normalize the extracted data into the shape your downstream system expects.
4
Routed into your ERP
Cleaned data flows into NetSuite, JD Edwards, SAP, or wherever your process needs it next — no manual keying.
Manual Transcription vs. OCR Automation
| Task | Manual Process | AtomIQ OCR Pipeline |
|---|---|---|
| Reading the document | Staff opens and reads each PDF or scan | Azure Document Intelligence extracts fields automatically |
| Entering line items | Manually typed into the ERP | Mapped and routed automatically |
| Transcription errors | Common on high line-item-count documents | Reduced — data pulled directly from source |
| Processing time | Minutes per document | Near-instant per document |
Works With How Your Documents Already Arrive
No need to change your intake process.
FTP
OpenAI Azure Blob Storage
Amazon S3
Document URL
Previous workflow task output
Frequently Asked Questions
How does AtomIQ read purchase order documents automatically?
AtomIQ uses Azure Document Intelligence to analyze purchase order and invoice documents, pulling structured field data such as line items, quantities, and totals directly from the document, whether it arrives as a PDF or scanned image.
Where can the source documents come from?
AtomIQ can pull documents from a file location such as FTP or cloud blob storage, or accept a document URL directly, depending on how your existing document intake process works.
What happens to the extracted data after OCR?
Extracted fields pass through an inline transformation step that maps and cleans the data before it’s routed into your ERP or downstream workflow, reducing manual re-entry and transcription errors.
Does this only work with invoices, or other document types too?
Azure Document Intelligence supports multiple document models, including invoices and purchase orders. AtomIQ’s workflow designer lets you configure which model applies to your document type.
Related AI Agent Workflows
See This Running on Your Own Documents
Bring a sample purchase order and we’ll show you the extraction live.