AppraisalAI — production lane
From MLS, field notes, and sketch to a signed MISMO 3.6 delivery package
AppraisalAI is the production lane of the DeepV platform. It ingests the appraiser's field evidence — MLS records, on-site notes, sketches, photos, supplements — and produces a UAD 3.6 form-fill, the MISMO 3.6 delivery package, a clean render-back, and a production-grade QC trace. The appraiser owns the final review and approval.
Three product lanes, one platform
Lane 2
FAR — forensic review
Forensic review on the full document package + chat + memory. Active build.
See FAR →Lane 3 — this page
AppraisalAI — production
Appraisal-production lane. MLS + notes + sketch → UAD 3.6 + MISMO delivery.
What AppraisalAI ingests
Field evidence is captured at the source. Each artifact is hashed, timestamped, and tagged against the field it supports — so the package stays reconcilable later.
MLS docs
Listing record, agent remarks, status history, DOM and price-change log, photo set, and tax assessor cross-walk. Provenance preserved per cited field.
Field notes
On-site inspection notes from the appraiser — measurements, condition observations, deferred maintenance, photo captions, and reviewer questions.
Sketch intake
Floor plan sketches and exterior measurements imported and tagged against the GLA grid. Stored as primary evidence, not a derived artifact.
From intake to delivery — six stages
Multi-source intake
MLS, field notes, sketches, supplements, and prior appraisal exports all land as first-class artifacts with hash + timestamp.
UAD 3.6 form-fill
Extracted data populates the UAD 3.6 form. Required / conditional / optional flags surface alongside missing-data callouts while the appraiser is still at the desk.
MISMO 3.6 package assembly
Schema-validated MISMO 3.6 XML is produced from the same data layer, including exhibit, sketch, and photo references.
Render-back to the appraiser
Structured payload renders back as a clean form view; the appraiser sees how the system read every field before submission.
Quality-control trace
A production-grade QC trace runs while the appraisal is being finalised — every parser pass, rule outcome, and field reconciliation pinned to a build version.
Appraiser review + approval
The appraiser owns the final review surface. They edit any field, attach addenda, accept or reject the QC findings, and sign the delivery package. DeepV does not sign on behalf of the appraiser.
Appraiser review and approval workflow
The appraiser owns the final review surface. AppraisalAI assembles the candidate package; the appraiser confirms, edits, attaches, and signs.
- 1AppraisalAI assembles the package from MLS + field notes + sketches + supplements.
- 2Appraiser opens the render-back view and confirms or edits each field.
- 3Appraiser accepts or overrides the QC trace findings; any override is captured in the audit trail.
- 4Appraiser attaches addenda, certifications, and any extraordinary assumptions.
- 5Appraiser signs the delivery package; the MISMO 3.6 export + render-back PDF + QC trace ship together.
Delivery outputs
One finalised AppraisalAI submission produces every artifact below in a single signed bundle.
MISMO 3.6 XML
Schema-validated structured appraisal payload, ready for investor delivery
Structured JSON
Same data, ergonomic for downstream review and audit consumers
Render-back PDF
Human-readable form view aligned to the source data
QC trace bundle
Production-grade audit pack (rule_pack_version, build id, field provenance)
Photo + sketch index
Field-tagged, hashed, timestamped, with MLS / EXIF cross-walk
Build posture
- AppraisalAI is an active build lane. Production submissions are scoped per engagement until the public production endpoint is opened.
- The same Section 15 (Lex Parsewell) extractor and Section 16 (Ken Guilfoyle) compliance brain that power AR / FAR review run in the production QC trace.
- Delivery-package signing, hash pinning, and SHA-pinned MISMO emissions are part of every QC trace.
- The appraiser remains the responsible party for the appraisal opinion; AppraisalAI is production support and form-fill, not an opinion-of-value source.
Delivery-readiness track
GSE delivery-readiness lives in AppraisalAI
AppraisalAI is the DeepV lane that produces the artefacts an investor-delivery programme evaluates: MISMO 3.6 XML, schema validation, render-back, photo and sketch provenance, the production QC trace, and the appraiser-signed delivery bundle. The DeepV review lanes (AR demo, FAR forensic review) are review and QC products — they do not produce delivery packages and are intentionally kept outside of any delivery-vendor approval track.
- MISMO 3.6 conformance is the spine.Every delivery is schema-validated against the published MISMO 3.6 reference model with a pinned ruleset.
- QC trace travels with every package.The same parser + rule-pack version identifiers used in AR / FAR review pin every production submission.
- The appraiser owns the opinion.AppraisalAI is form-fill and delivery support. The signed appraisal opinion is the appraiser’s; DeepV does not act as an appraiser.
- Approval status is scoped per engagement.Where a customer requires alignment with a specific investor delivery programme, that alignment is worked into the AppraisalAI scoping call — not claimed on a marketing page.
Scope an AppraisalAI production engagement
Bring the form set you produce and the delivery contract you ship against. We will scope an AppraisalAI integration that lands the MISMO 3.6 package, the render-back PDF, and the QC trace in one signed bundle that the appraiser owns end-to-end.