Automated Appraisal Review · FNMA & FHLMC

A 60-page appraisal, reviewed in minutes. Zero humans in the loop.

Residential appraisals arrive as 35-to-70-page scanned PDFs — the Fannie Mae form plus dozens of interior, exterior, comparable, map, license and E&O photos. The engine extracts 450+ data points and every image, runs the full FNMA/FHLMC rulebook, and hands the Collateral Underwriter a cleared, exportable review — with no human-in-the-loop, even on low-resolution scans.

Appraisal Hero Banner
450
Data items extracted per report
210
Rules in the FNMA/FHLMC library
3
Parallel analysis engines
35–70
Report pages handled per file
0
Human-in-the-loop steps
Why it's different

Built to do what other tools can't

The hard problems in appraisal review are messy scans, unbounded appraiser prose, and dozens of photos. This engine handles all three at production scale.

Reads low-res scans

Extracts data and images accurately from poor-quality scanned PDFs — the exact input where competing tools break down.

True computer vision on photos

Runs real image analysis on every photo and judges condition and defects from the pixels themselves — it does not just trust what the appraiser wrote.

Reads pages of comments

Scans the appraiser's free-text throughout the report to surface any observation — anywhere across dozens of pages — that could affect the appraised value.

Reads companion documents

Ingests the Appraisal SSR and the purchase contract too, so the appraisal is reviewed in context rather than in isolation.

Pulls live market data

Fetches third-party MLS and AVM data to run cross-checks the appraisal alone can't support — including occupancy-fraud detection.

Zero human-in-the-loop

No HIL step anywhere in the pipeline. The review is fully deterministic end to end — nothing waits on a person.

Infinitely scalable

With no human bottleneck, throughput is limited only by compute — scale horizontally on AWS with consistent TAT.

Minutes, not hours

A 60-page report is reviewed in minutes, collapsing what is otherwise a long manual stare-and-compare.

Rules you can change fast

Highly customizable engine built to FNMA and FHLMC guidelines — new rules are added quickly as policy shifts.

Rules by report section

Where the 210 checks apply across the appraisal

Fannie Mae forms the engine reads

The right form per property type — plus the 1004MC market-conditions addendum
Form 1004
Single Family (URAR)
Form 1073
Condominium
Form 1004C
Manufactured Home
Form 1025
2-4 Unit / Small Income
Form 1004MC
Market Conditions Addendum
Form 1004D
Update / Completion
Beyond the appraisal

Cross-checks other documents and live market data

The engine doesn't just read the appraisal. It reconciles it against the SSR and purchase contract, and pulls third-party MLS/AVM data to run rules like occupancy-fraud detection.

Sources reconciled
Appraisal Report
The scanned form + all photos & exhibits
Appraisal SSR
Submission Summary Report from UCDP
Purchase Contract
Price, parties, terms & dates
MLS
Listing history & status
AVM
Independent value & market signals
↓

Contradictions surface automatically

Value, occupancy, ownership, dates and sale terms are compared across every source — mismatches become findings before a human ever opens the file.

Findings

Every check returns a graded result

Boolean checks pass or fail; photo-based condition checks return a severity so the underwriter can triage fast.

Result grades
Pass Fail N/A Defect severity → None Low Medium High

142 checks fired on a 60-page report

A live sample: a Conventional · Purchase · Single-Family file on Form 1004 activated 142 of the 210 library rules — data, text and image — reviewed end to end with no human intervention.

Outputs & integration

Delivered everywhere the file needs to go

The Collateral Underwriter works in the portal or in Encompass — the engine keeps both in sync and feeds downstream systems.

Web portal

Interactive checklist — the underwriter clears failed rules online and exports the final review straight to Encompass.

Encompass eFolder

The review PDF is uploaded automatically, so the file is complete without any manual filing.

Automated conditions

Underwriting conditions are created directly inside Encompass from the review's findings.

MISMO XML

A machine-readable appraisal is generated for any downstream system that needs structured data.

The point: collapse Collateral Underwriter review time

Fully deterministic and human-free, the service scales horizontally on AWS with consistent turn-around — driving lower cost per file, higher throughput, and uniform quality. New FNMA/FHLMC rules are added quickly as guidelines change.