SITUS
PCA. ESA. Appraisal. From one drive-by.
Vision-AI captures collateral, environmental, and condition diligence in 48 hours. For lenders, receivers, and institutional CRE diligence teams.
Photos scattered in folders. Costs buried in spreadsheets. Context locked in engineers’ heads.
SITUS turns every assessment into structured, auditable intelligence.
Watch the walkthrough →Federal data, auto-fetched
- Pulling zoning by hand
- Re-checking flood maps every time
- Stale environmental records
SITUS pulls FEMA, EPA, NOAA, NRCS, USGS with citations on every claim.
AIExtracts and reconciles federal datasets into one cited property record.
SourcesFEMA → EPA → NOAA → NRCS → USGSAny file, any source
- Photos buried in folders
- Hand-logging components
- Forgetting which evidence proves what
Reads photos, drone, LiDAR, thermal, PDFs. Autofills the checklist, reviewer keeps the call.
AIReads photos, drones, LiDAR, and PDFs. Autofills component, condition, observation, and action.
InputsPhotos · Drone · LiDAR · Thermal · PDFsFindings priced as you go
- Findings in one tool, costs in another
- No link from finding to dollar
- Reserve tables rebuilt from scratch
Every finding maps to your cost library. One click builds the priced reserve.
AIMaps each finding to your cost library. Builds the priced reserve in one pass.
FlowFinding → Cost → ReserveAudit-ready in hours, not weeks
- Days reformatting templates
- Manual cross-references
- No trail back to the evidence
Full PCAs or Quick Response. Every data point traces back to a finding and a photo.
AIDrafts the narrative from approved findings. Reviewer edits and signs off.
OutputPCA · Quick Response · Provenance chainAssessments die in static reports.
AI workflows that capture, price, and prove every line.
Assessment
Harbor Common Ctr
Harbor Common Ctr
Concrete Surfaces
Common Area Amenities
Current Deliverable
Assessment
Harbor Common Ctr
Harbor Common Ctr
Concrete Surfaces
Common Area Amenities
Current Deliverable
One Project. Start to Finish.
The kind of review FDIC receivership, lenders, and due-diligence firms need – in under 3½ minutes.
What changes when SITUS runs the workflow
Same people. Same properties. Entirely different output discipline.
Google the address. Check county records manually. Hope someone remembers what happened last time.
Property Scan, hazard APIs, and satellite recon run in parallel. flood zones, EPA sites, soil, seismic, wind data returned in seconds.
Google the address. Check county records manually. Hope someone remembers what happened last time.
With SITUSProperty Scan, hazard APIs, and satellite recon run in parallel. flood zones, EPA sites, soil, seismic, wind data returned in seconds.
Uploaded to a shared drive. Manually dragged into report sections. or skipped with a note that says "see appendix."
Drive-By AI analyzes uploads during the site visit. binds each photo to a checklist item with observations and evidence IDs.
Uploaded to a shared drive. Manually dragged into report sections. or skipped with a note that says "see appendix."
With SITUSDrive-By AI analyzes uploads during the site visit. binds each photo to a checklist item with observations and evidence IDs.
Reviewer picks a condition rating from a dropdown. No connection to what the cost table says about timing.
Condition auto-derives from findings: Replace → Poor, Repair → Fair, Monitor → Good. Evidence drives the rating, not opinion.
Reviewer picks a condition rating from a dropdown. No connection to what the cost table says about timing.
With SITUSCondition auto-derives from findings: Replace → Poor, Repair → Fair, Monitor → Good. Evidence drives the rating, not opinion.
Entered in a cost table that doesn't talk to the condition field. Both filled in separately, by hand.
Cost timing and condition are validated against each other. Priority-timing mismatches are surfaced and tracked until corrected.
Entered in a cost table that doesn't talk to the condition field. Both filled in separately, by hand.
With SITUSCost timing and condition are validated against each other. Priority-timing mismatches are surfaced and tracked until corrected.
Assembled inside a template that enforces nothing. Quality depends entirely on who's writing that day.
Generated from confirmed findings. Narrative, cost table, photo appendix. assembled, not written. One template, zero variance.
Assembled inside a template that enforces nothing. Quality depends entirely on who's writing that day.
With SITUSGenerated from confirmed findings. Narrative, cost table, photo appendix. assembled, not written. One template, zero variance.
Opens the finished PDF and starts over. Three exports to get the page breaks right.
Reviews staged findings inline. Mismatches are flagged at entry and tracked. caught before export, not after.
Opens the finished PDF and starts over. Three exports to get the page breaks right.
With SITUSReviews staged findings inline. Mismatches are flagged at entry and tracked. caught before export, not after.
Waits for the complete report. The reserve number isn't separated from the narrative. you can't get one without the other.
Quick Response ships the reserve number the day findings are confirmed. Full narrative follows on its own timeline.
Waits for the complete report. The reserve number isn't separated from the narrative. you can't get one without the other.
With SITUSQuick Response ships the reserve number the day findings are confirmed. Full narrative follows on its own timeline.
Drive-By
SeaGoat's vision-AI engine.
Reads any file. Autofills the checklist. The reviewer keeps the call.
What's in every Drive-By
Three things. Every time.
Identification
Components and systems Drive-By reports on:
- Site · stormwater · paving · landscape
- Building Exterior · roof · walls · windows + doors
- Mechanical · HVAC · electrical · plumbing
Description
Plain-language condition write-up per component, in ASTM E2018-24 phrasing. Every claim tied to its source. Photos, drawings, and imagery from multiple sources.
Cost Tables
Capital expenditures by component and timing horizon. Repair, Replace, or Monitor, projected from building history and comparable properties.
Want interior coverage too?
Drive-By can't see inside. Interior systems can be approximated separately.from industry standards, building age, and comparable properties.
ASTM E2018-aligned
Component, condition, and observation in the standard phrasing every PCA needs.
Explore SITUS
Click through every step. One engine.
Walk the Central Tower assessment from Research through Report. Every tab is a real stage of the workflow.
#PCA-22 · 450 Commerce St, Austin TX#PCA-2024 · 450 Commerce St, Austin TX
Property Briefing
Property
Systems
Central Tower is a 5-story Class B office building with approximately 127,000 SF situated on a 2.9-acre urban site. Constructed in 1974. Flat/low-slope built-up roof with multiple sections – age exceeds typical 20–25 yr expected useful life. Rooftop HVAC package units and AHUs present. Surface asphalt parking. Facade: glass and precast concrete. HVAC systems likely beyond serviceable life. Elevator count to be confirmed on site.
Public Records · Regrid API · Satellite Imagery · County Appraisal District
Zoning & Land Use
Dimensional
Permitted Uses
Zoning Code · Regrid API
Hazards
Flood Zone
FEMA APINon-SFHA
Minimal flood risk
Low
Wind Zone
ASCE APIZone II
130 mph – design required
Moderate
Seismic Zone
USGS APISDC A
Low seismic activity
Low
Radon Zone
EPAZone 3
Low potential
Low
Climate Zone
NOAA APIZone 2A
Hot–Humid
IECC · NWS
Environmental Flags
EPA APIFlagged facilities nearby
Phase I ESA scope review recommended
Review recommended
Soil / Foundation
NRCS APIUrban fill
Gentle slope · Variable drainage
USDA Soil Data
The brief is built before you leave the office.
Six federal sources, county GIS, and grounded AI search, queried in parallel from one coordinate. Every fact links back to its source. The inspector drives out with the building already documented.
Eight tabs become one query.
Federal records, municipal data, satellite imagery, listing sites, pulled in parallel, deduplicated, and synthesized in under a minute. No copy-paste. No tab juggling. No regulated site half a mile away slipping through.
Each source returns its own evidence. Each finding carries a citation. Every claim ships with a record behind it.
↳ Zone X · 0.2% annual
↳ 0 regulated within radius
↳ Zone 2A · 115 mph design
↳ Site Class D · low risk
↳ Clay 32% · moderate swell
↳ CBD · 2.9 ac
The subject property is Central Tower, a 5-story Class B office building in downtown Austin, TX. GBA is approximately 127,000 SF on a 2.9-acre urban site. Constructed in 1974, the building presents typical mid-century commercial construction with original roof and core mechanical systems. Stabilized occupancy is approximately 78%.
Roof is flat/low-slope built-up membrane. age exceeds the typical 20-25 year expected useful life and warrants close attention on site. Rooftop HVAC consists of package units and AHUs, several likely past serviceable life. Zoning is CBD.
Auto-generated. Confirm on-site before sign-off.
Every claim shows where it came from.
The brief reads like a paragraph and ships like a citation chain. Every fact, from square footage to zoning to flood zone, links back to the record it came from.
Six months later, when an underwriter or auditor follows up, the source is already on file. No reconstructing a paper trail from memory.
Every fact carries its source.
FEMA flood maps, EPA environmental records, USGS seismic, NOAA wind, NRCS soil, Regrid parcels, county data. Every claim links back to the record it came from. If an underwriter or auditor follows up months later, the citation chain is already in place.
AI proposes. Reviewers prove.
Drive-By AI reads each photo as it's uploaded, proposing the component, the condition, and the observation in ASTM E2018-24 phrasing. The reviewer confirms or overrides. Every change logs name and rationale.
Every override logs the reviewer.
Drive-By AI proposes. The reviewer confirms or overrides, and the change carries a name, a timestamp, and a rationale into the report. No reconstructing decisions from memory.
The same chain that lets the AI propose lets the firm prove who decided what, six months from now, when it matters.
The inspector walks in with a checklist.
Every red flag from Research becomes an investigation action, numbered, traceable to the source that triggered it. The reviewer marks them off as the walk progresses.
No more “what did we say to look at?” the morning of the visit. The list is on the iPad before the truck pulls out.
Inspect roof membrane at parapet and low points
↳ USGS · satellite history
Confirm elevator count, document cab condition
↳ TCAD · permit gap 2001
Verify HVAC RTU make, model, age
↳ Prior PCA spec sheet
Inspect south facade caulk deterioration
↳ FEMA wind zone
Check sidewalk and ADA compliance
↳ Web search · code
An AI that won’t make things up.
Per-image triage following ASTM E2018-24 phrasing. The model classifies hazard, defect, and component, and ties every finding to its source photo. It is explicitly forbidden from inventing dollar estimates, assigning a final building condition, or labeling normal joints as defects.
A capital plan the underwriter doesn’t argue with.
Every finding feeds a structured cost record. Every cost ties to your firm's library. Every change recomputes the schedule, the reserve total, and the underwriting comparison, automatically.
Every finding becomes a line on the schedule.
Each finding carries a remaining useful life (RUL = EUL − Age), a replacement year, and per-line inflation. The 12-year schedule generates from that. Immediate, Short-Term, Long-Term, Total Reserve.
Edit a condition. Swap an action. The schedule recomputes on the spot. No more stale PDFs.
4 events
6 events
4 events
Per S&P Underwriting Guidelines for Structured Finance
How it stacks up against the underwriter.
SITUS compares every reserve against industry benchmarks. S&P Underwriting Guidelines, lender requirements, internal thresholds. The computed-vs-expected delta sits in the report.
If a number is off, you see it before submission, not after the underwriter calls.
Every cost cited to your firm’s library.
No third-party royalty, no black-box index. Costs come from the cost library you control, or a vendor quote, or a reviewer override with rationale logged. When the underwriter pushes back, the source is already on file.
Numbers to the lender. Story to the file.
Two deliverables from the same evidence chain. Quick Response ships first, with the cost table and reserve schedule lenders need to close the loan. The full narrative follows the same week, with photo appendix and compliance language already assembled.
Sample Deliverables
01Quick Response Report
Quick Response
Riverside Commons · 4820 Harbor View Blvd, Austin TX · March 19, 2026
Opinions of Cost
| Horizon | Items | Cost |
|---|---|---|
| Immediate | 1 | $0 |
| Short-Term | 2 | $0 |
| Long-Term | 1 | $0 |
| TOTAL | 4 | $0 |
Underwriting Guidelines: Computed $381,026/yr vs. S&P guideline $317,500/yr ($2.50/Unit). Delta: 20% above guideline.
Cost Detail
| Item | Driver | RUL | Total |
|---|---|---|---|
| Roof Membrane Replacement | Roofing Systems | 0 | $0 |
| HVAC System Replacement | HVAC Equipment | 0 | $0 |
| Elevator Modernization | Vertical Transport | 5 | $0 |
| Sidewalk & Curb Repair | Exterior Hardscape | 10 | $0 |
The reserve number
ships before the
narrative is done.
Lenders open the cost table first. Quick Response delivers exactly that. reserve schedule, opinions of cost, immediate-vs-reserve timing. the moment reviewer sign-off is complete. No paragraph edits holding up the number that closes the loan.
Cost-first output
Reserve schedule, cost table, underwriting comparison. From confirmed findings only.
02Full Field Report
The story behind
the numbers.
Cover, Project Summary, Condition Matrix, Capital Reserve Schedule, Photo Appendix. assembled from confirmed findings, narrated automatically, formatted to your firm's template. No copy-paste. No reformatting after every revision.
Auto-assembled sections
Cover, Summary, Condition Matrix, Reserve Schedule, Photo Appendix. Your firm’s logo, your firm’s language, your firm’s template. SeaGoat fills the pages; the brand on them stays yours.
Property Condition Consultants, Inc.
Due Diligence · Engineering · Risk
Property Condition Assessment Report
Riverside Commons
4820 Harbor View Blvd, Austin TX 78701
Prepared For
Meridian Capital Group
Project No: PCA-2026-001
Prepared By
Property Condition Consultants, Inc.
Property Condition Assessment
Property Condition Assessment · Sample Output
Demonstration Only. Generated by SITUS
Your firm’s branding on every cover.
Your logo, your colors, your title block, your transmittal letter. SeaGoat doesn’t sign your work. Every report comes out looking like yours because it is yours. Cushman-branded, BBG-branded, your-firm-branded.
Kickoff to lender's desk in 48 hours. Full narrative on the file the same week.
One engine. Many Packs. Yours forever.
Run multifamily PCAs today. Add HUD CNAs, ESAs, or Progress inspections next quarter. Same engine, same library, same audit trail. Your Pack carries your firm's IP. Editable, audit-tracked, exportable on demand.
SITUS Core
Engine, schemas, ASTM-aligned rules, audit trail.
Your Pack
Your firm's cost library, checklists, evidence templates, QC standards. Edit anything. Audit-tracked.
Project Layer
Per-project overrides, vendor quotes, override rationale. Standards enforce at export. Never poisons your Pack.
Configured for every asset class your firm runs.
Each asset class carries its own checklist template and component defaults. fully editable in your Pack.
Distribution & logistics
Apartment building
Commercial campus · Solar PV
Each project type drives report depth and AI analysis.
Tag every project at kickoff. Report depth and AI depth adjust automatically. Standard PCA for lender review, expanded analysis for equity diligence. Debt and Equity are seeded; add Construction, Acquisition, Refi, or your own labels.
Build a checklist for any Assessment Type.
PCA, ESA, Appraisal, or your firm's specialized engagement. Every Assessment Type carries its own checklist with its own components, EUL defaults, and evidence requirements. New industry standard? Build it once in your Pack. Custom client requirement? Add it as a one-off.
Your firm's costs drive every reserve number.
Your firm's negotiated rates, market data, and vendor quotes. Loaded once, used on every project. No third-party royalty. No black-box index. Every change logs who, when, and why.
PCA
Property Condition Assessment
The kinds of work your firm runs.
Each Assessment Type drives its own editable checklist and AI depth. Your firm names them. PCA, ESA, Progress Inspection, MAI Appraisal, or your own internal labels. Add, rename, or rewire checklists in your Pack at any time.
Your Pack is your firm's IP. Yours forever. Audit-tracked. Exportable on demand.
Ask anything.
Answers that know your building.
ChatGPT doesn't know your project. Ask SITUS does. It already has the address, the year built, the satellite history, and every finding from the file. Same model power. Applied to your file, not a blank slate.
Stop pasting context. Ask SITUS already knows the building.
Ask the same question you'd ask ChatGPT. The answer is specific to your property because the project file is already loaded. Address, zoning, construction type, satellite imagery, research findings, site conditions, cost data. No copy-paste. No re-prompting.
Live web search when codes and permits matter.
Codes change quarterly. Permits update weekly. SITUS pulls live so the answer matches reality, not what was true two years ago. Every source cited.
This isn't something you build in‑house.
SITUS is built at the intersection of two rare disciplines: deep industry expertise from senior CRE diligence practitioners, and modern AI engineering from production deployments of safe, governed LLM systems. The result is forty thousand lines of proprietary domain logic. Every threshold defended, every edge case named, every output traceable. A specimen from inside the engine:
threshold = f( human_overrides, t )
Confidence thresholds are not constants. They are recalibrated against incoming reviewer corrections every release. The curve only flattens against corrections at scale, not synthetic benchmarks.
corpus = mistakes
+ corrections
+ rationalesThe training data that makes the engine work is not the right answers. It is the wrong ones, the human corrections, and the rationale behind them. Captured from real reviewer overrides, not licensed from a vendor.
ai_engineer ∩ senior_pca ≈ ∅
Two scarce hires that rarely overlap. Production AI engineers and senior CRE practitioners. The intersection is statistically empty and expensive when found.
claim → evidence
→ reviewer
→ ledgerEvery assertion carries a verifiable lineage. Source ID, evidence IDs, reviewer, override rationale, timestamp. A provenance state machine built into the schema, not bolted on.
agree(model_a, model_b, model_c) || reviewer.hold
Multi-model agreement with disagreement-triggered hold. Foundation models vote. When they conflict, the call routes to a qualified human. Responsible-AI patterns baked in from day one.
SITUS ⊂ {SITUS, GRID, ATLAS}SITUS is one engine on a fabric. A firm building "a SITUS" in-house ships one tool. We ship the platform that compounds with every visit.
None of this is a feature list. It is the moat.
Defensible by design.
SITUS is built for environments where outputs must hold up years later. In audits, legal reviews, and lender scrutiny.
Every claim carries its receipt.
A source ID, evidence IDs, a timestamp, a version, a review status, and an override rationale when a human corrects the engine. Provenance states are explicit: Verified means modern evidence plus reviewer approval; Resurrected means pulled from archive, unverified until confirmed; Derived means inferred from upstream facts and required to show lineage.
When sources disagree, the engine knows who wins.
Construction documents rank highest. As-builts, specs, permits. Below them, physical assessment evidence. Then county and municipal records. Then third-party prior reports. Client-provided spreadsheets sit lowest. Overrides require rationale plus reviewer approval. Same-level conflicts flag for human review. “Unknown” is always a valid answer. The engine never forces a guess.
Models suggest. The system validates. Humans decide.
AI proposes field values, conditions, and costs. Humans confirm or override every one of them. Each override requires a written rationale before it saves. Sign-off is reviewer-specific and timestamped, never a blanket approval. Nothing reaches the report without explicit human staging, and every staging action lives in the audit log, ready for legal review or lender scrutiny.
Prior reports ingest as history, not as truth.
Documents chunk by type. Tables, narrative, bullet lists. Provenance preserves page and section references where the source allows. Ingested data defaults to Unverified until a reviewer confirms the finding. Nothing auto-promotes. When modern evidence contradicts a legacy finding, the old one becomes Superseded, never deleted. The audit trail keeps both.
SITUS is the entry point. Intelligence compounds.
Every verified output feeds GRID's asset memory. ATLAS turns that memory into portfolio intelligence.
Assessment Layer
Captures the assessment. Every approved PCA, ESA, Drive-By, and reserve schedule flows into GRID as part of the property record.
The more you use SITUS, the smarter the whole system gets.
Deploy the engine.
We’ll map your deliverables and stand up a workspace.