Authenticate to the Ocrolus API using an API key and create a Book (document collection container) for the borrower's loan file
Upload income documents (pay stubs, W-2s, 1099s, tax returns) to the Book via the document upload endpoint, specifying the document_type for each file
Poll or use the callback webhook to wait for the Book to reach COMPLETE status, indicating all documents have been classified and extracted
Call the Analyze for Mortgage endpoint on the completed Book, specifying the applicable agency guidelines (Fannie Mae, Freddie Mac, FHA, VA, or USDA) for income calculation
Retrieve the structured income analysis response, which includes calculated qualifying income per income stream, supporting document evidence references, and any calculation discrepancies flagged
Integrate the Ocrolus Inspect findings into the LOS by mapping extracted income values to the corresponding loan fields, using the document-level confidence scores to flag items requiring underwriter review
Known gotchas
Ocrolus supports over 1,600 document types, but the income calculation logic is sensitive to document quality; blurry or low-resolution scans often extract raw data correctly but fail the calculation rules, requiring manual review rather than automated override
Agency income calculation guidelines (Fannie Mae vs. FHA vs. VA) produce materially different qualifying income figures from the same documents; always explicitly specify the guideline set rather than relying on any default
The Encompass integration syncs Ocrolus findings directly to loan fields, but field mapping is version-specific; after Encompass releases updates, re-validate that the field IDs in the Ocrolus-to-Encompass mapping still correspond to the correct loan data elements
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp