Processing API

Changelog

Rowland Processing API Version History

v2.15.0 · May 2026 (Latest)

This release improves extraction reliability and cleans up how repeated data is returned.

What's improved

Per-track retry handling. Document digitization and extraction now retry independently per track, so a transient failure in one stage no longer forces the whole job to restart.
Latest royalty rate now returned. Fixed an issue where document extractions could return a stale royalty rate instead of the most recent one.
Tract counts instead of tract lists. Extractions now return a count of tracts rather than enumerating each tract individually, producing smaller, more predictable payloads.
Security and dependency updates applied to the Processing API

What this means for integrators

Fewer full-job retries and more resilient extractions
Accurate royalty rate values without needing manual reconciliation
Heads up, shape change: if your integration reads individual tract entries from extraction results, update it to consume the new tract count field instead.

Previous Releases

v2.14.0 · May 2026

Improved OCR reliability, extraction accuracy, and visibility into document processing status.

Fixed OCR workflow reporting so failed OCR runs no longer appear successful
Clearer real-time OCR and document-processing status visibility, including better handling for partial-success cases
Extraction workflows prefer cleaner text inputs while preserving fallbacks when needed
Improved consolidation for large and complex documents, including better handling of cross-references
Reduced truncation and retry issues during Gemini-powered PDF processing
Improved source extraction for long text values
More accurate document highlighting in previews
Reduced application log noise from health check and heartbeat traffic

v2.13.0 · May 2026

Improved upload and retry request handling behind authenticated app-managed routes
Improved lease extraction so named clauses can be captured more directly for downstream use

v2.10.0 · April 15, 2026

This release focuses on making document processing more reliable and easier to trust as well as consolidating the APIs under a single URL.

What's improved

Better reliability when processing large or complex documents
Smoother handling of document conversion steps, including searchable PDFs
Faster, more accurate progress updates while documents are being processed
Ongoing improvements to the Processing API platform to make the service more stable
Behind-the-scenes security and infrastructure updates

What this means for users

Fewer processing interruptions
Better visibility into document progress
A more consistent experience overall

Most of the work in this release is behind the scenes, so the biggest benefit is improved stability rather than major changes to how the product looks or works.

Documents

v2.8.0 · April 6, 2026

Improved handling of malformed AI responses during document processing
Fixed memory issues when processing large scanned PDFs

v2.7.0 · March 30, 2026

Searchable text layers for scanned PDFs
Extraction results with page numbers and source text
Duplicate file detection via content hashing

v2.6.0 · March 2026

Consolidating v0.2.0–v0.6.0

Expected document type parameter for targeted extraction
Automatic correction of rotated PDF pages
Upgraded to Gemini 2.5 Flash-Lite for faster processing
AI-assisted arithmetic validation for extracted numerical data
Folder-scoped document operations
Batch processing with parallel document uploads
Configurable output token limits for extraction responses
Document table pagination and filtering

v0.1.0 · September 12, 2025

Initial release.

Workflows