SITUS
The Super-Inspector for construction due diligence.
SITUS runs the workflow—not just assists it.
Assessments Die in
Static Reports.
Most construction intelligence stays trapped: photos scattered in folders, costs buried in spreadsheets, and context locked in engineers' heads. When teams change or projects hand off, the knowledge disappears. Clients get conflicting data when ESA and PCA reports don't sync.
SITUS turns every assessment into structured, auditable intelligence—so findings become costs, costs become schedules, and schedules sync to your institutional memory. Unified Salient Facts: core asset data entered once, synced across PCA and ESA.
What changes when SITUS runs the workflow
Same people. Same properties. Entirely different output discipline.
Uploaded to the platform. Manually dragged into sections — or skipped with a note that says "see appendix."
Entered in a cost table that doesn't talk to the condition field. Both filled in separately, by hand.
Assembled inside a template that enforces nothing. Quality depends entirely on who's writing that day.
Opens the finished PDF and starts over. Three exports to get the page breaks right.
Waits for the complete report. The reserve number isn't separated from the narrative — you can't get one without the other.
Prior assessments found manually. Asset history lives in email threads and a shared drive no one maintains.
Drive-By AI attaches them to checklist items before anyone goes onsite. Evidence is bound, not filed.
ASTM-driven: cost timing determines condition. The table and the condition rating are the same source — no contradictions.
Generated from confirmed findings. Narrative, cost table, photo appendix — assembled, not written. One template, zero variance.
Reviews staged findings only. Mismatches block save — caught at entry, not after three PDFs.
Quick Response ships the reserve number the day findings are confirmed. Full narrative follows on its own timeline.
Prior reports link to the property automatically. Asset history is structured, searchable, and always current.
The Engine Runs the Workflow
Old way: Inspectors manually assemble photos, tables, and narrative sections.
SITUS way: The engine drives the process by default.
INGEST
Photos, PDFs, legacy reports. Evidence becomes structured observations.
STRUCTURE
Proposed findings with condition tags. Bound to costing logic and reserve schedules.
ASSEMBLE
Findings, cost tables, report sections. Generated from authoritative records.
GOVERN
Human review + approval. Every override logged with rationale.
Result: Less cognitive load. Fewer review failures. Consistent outputs across teams.
Evidence-First Logic: From Intake to Output
Guardrails That Prevent Bad Outputs
SITUS enforces evidence discipline and keeps a clean chain of custody—from intake → finding → cost → narrative. It eliminates the "data gap" between the field and the final report.
Evidence Binding
Poor condition requires photo evidence. Missing access triggers "Limited Observation" language. Every finding traces back to source documents.
Cost Logic
Repair/Replace requires quantity + unit cost. Cost binds to condition via deterministic rules. Reserve schedules generated from authoritative records.
Liability Gating
Human review + approval required. Every override logged with rationale. Permanent audit trail for defensibility in legal reviews and lender scrutiny.
No guessing. No improvisation. No downstream surprises.
Installed Packs
SITUS adapts to your workflow. Select packs to inherit validation rules, evidence requirements, and reporting standards.
Property Condition
PCAEnvironmental
ESAProgress Monitoring
CONSTRUCTIONNeed a custom workflow? We build it with you.
Apply for Partner TrackSystem Workflow Demo
Click each tab to run that workflow. Research handles recon and pre-site actions, Site handles checklist + finding staging, Findings shows costed items, and Reports generates downloadable output.
Checklist
Property name
Central Tower
Elevator count
Current: 28
Floors served
Current: 55
Auto-fill Research fields with ✦ AI Research, Satellite Property Scan and Google Earth.
Observations
Elevator systems show aging modernization indicators from historical records and prior reports.
Findings
Potential modernization requirement detected; verify controls, cabs, and machine room condition on site.
The Findings Record: Structured Evidence
Every finding becomes a structured record with evidence binding, cost logic, and review status. Guardrails prevent bad outputs.
ID: ALG-2026-001-042
Roof Membrane (Building A)
Poor
Replace
120 SF
Roof Membrane Replacement
$45,200
Immediate
stage_2026_001
[042, 043, 044, doc_001]
J. Doe (PE)
None
Guardrails enforce quality: Poor requires evidence. Repair requires quantity + cost.
Evidence to Decision, End to End
SITUS is built around one operational spine: recon evidence turns into mapped checklist actions, then deterministic cost logic and report output.
RESEARCH
Pre-site intelligence: Google Earth Timeline AI, Property Scan, and AI Research run together.
Timeline screenshots, public records, satellite imagery, and pre-site observations flow into structured recon. Observations generate on-site actions.
SITE
On-site evidence capture with Drive-By AI and checklist completion.
Drive-By AI auto-assigns files to checklist items. Reviewer confirms condition, quantities, and evidence per component. Staged findings flow to Findings.
FINDINGS
Staged items become structured findings with deterministic cost logic.
Repair/replace conditions bind to quantity, cost item, and reserve timing. Reviewer confirms each finding before it can appear in a report.
REPORTS
Report sections generate from confirmed findings with traceable citations.
Each claim traces to checklist data, evidence IDs, and reviewer sign-off. Reports require human approval before delivery.
Every output is traceable. Every decision is defensible.
You arrive on site
already knowing
what to look for.
Before your inspector sets foot on the property, STAGE's Research Engine runs three parallel scans and hands them a pre-site brief: what changed on the property over time, what the county says about it, and what to investigate when they get there.
The brief generates on-site Actions — specific, numbered investigation items that map directly to checklist fields. Reviewers don't assign work on arrival. It's already assigned.
Google Earth Timeline AI
Analyzes satellite imagery across years. Flags roof changes, additions, demolitions, and site modifications that don't appear in permits.
Property Scan
Cross-references county records, municipal data, and ownership history. Surfaces discrepancies before the visit — year built, last sale, zoning conflicts.
AI Research
Analyzes uploaded documents, images, and satellite data to generate proposed checklist fields, observations, and condition flags — all cached to the source, not re-run.
Property
1247 Oleander Ave · 1983 · 48 Units
Roof surface color change detected on Building C. Possible membrane replacement without permit on file.
Year built discrepancy: client states 1985, county shows 1983. Affects ASTM useful life calculations.
Proposed: Parking lot — Poor condition. Staining at building base. Downspout displacement at NW corner.
Generated On-Site Actions
Verify Building C roof — confirm replacement scope and permit status
Confirm year built with management — affects reserve schedule
Inspect NW downspout and base staining — moisture intrusion risk
Built for Flexibility, Not Forks
SITUS Core is the shared workflow foundation—reusable across construction domains. Workflow Packs configure client-specific behavior without fragmenting the codebase.
Packs are versioned YAML/JSON. base_pack → tenant_pack → project_override. Lower levels can only restrict—never expand. Invalid pack blocks project creation.
STAGE CORE
Reusable across PCA, ESA, Progress. Single codebase.
- →Evidence ingestion + staging
- →Findings object model
- →Findings + reserve logic
- →Report generator
- →Audit + provenance
WORKFLOW PACKS
Pack controls: rules, terminology, validation, templates.
- →Validation rules (e.g. debt vs equity)
- →Terminology (e.g. PCA, ESA, Progress, Appraisals)
- →Cost libraries
- →Templates + export formats
- →Domain-specific prompts
YOUR CONFIGURATION
Opt-in per pack. Nothing turns on without approval.
- →Enabled per tenant via entitlements
- →Partner proprietary logic lives in custom packs (e.g. PCA Pack)
- →Need PCA-style deliverables? There’s a pack for that.
- →Progress monitoring, appraisals, ESA—each use case can have its own pack.
What Packs Actually Control
- Validation rules: Poor requires photo. Repair requires quantity + cost. Pack-configurable thresholds (e.g. debt vs equity sensitivity).
- Terminology + language: Condition labels (e.g. Immediate vs Short-Term vs Routine). Action labels (e.g. R&M vs Replace). Mode selector (e.g. PCA, ESA, Progress, Appraisals).
- Cost + reporting: Cost libraries, EUL/RUL sources, inflation model. Report sections, required elements (FOIA, PSQ, ADA screening), output formats (PDF, CSV, photo appendix).
- SITUS → GRID sync: On approval, findings + evidence_ids + cost table + salient facts + override rationales sync to institutional memory. Append-only. Each sync creates an immutable snap.
Nothing turns on without approval—capabilities are opt-in per pack. Invalid pack blocks project creation. All packs validated against schema on load; schema changes require a migration path.
Quick Response
1247 Oleander Ave — PCA
$187,400
Immediate
$342,000
Year 1–3
$891,200
Year 4–10
Top Flags
Roof Membrane — Bldg C
Replace · Immediate
$45,200
Parking Lot — North Section
Replace · Year 1
$68,400
Elevator Modernization
Replace · Year 2
$112,000
Full narrative in progress
14 findings · 100% reviewedThe reserve number
ships before the
narrative is done.
Lenders don't need the prose. They need the cost table, reserve schedule, and key flags — and they need it fast. STAGE's Quick Response delivers a cost-first output the moment findings are confirmed, while the full narrative is still being assembled.
No waiting on writing. No holding the reserve number hostage to paragraph edits. The output that closes the loan ships first — the report follows.
Cost-first output
Reserve schedule, cost table, timing flags — from confirmed findings only.
Ships immediately
Generated the moment reviewer sign-off is complete. No assembly required.
Full report follows
Narrative, photo appendix, and compliance language come after — same evidence base.
Built for Teams That Can't Afford to Be Wrong
Per-report subcontractor costs mostly cover formatting work. SITUS generates the deliverable from confirmed findings — reviewers review, not rebuild.
Cross-referencing PCAs by hand against your own data wastes deal time. Structured findings let you compare condition, cost, and timing across assets in one place.
Defensible reserve numbers require a traceable evidence chain. Every finding connects to source, photo, and sign-off. Numbers that hold up under review.
Reviewing a finished PDF from scratch takes hours. Reviewers on SITUS see only staged findings. Mismatches block save — they're caught at entry, not in review.
Built for Liability-Heavy Work
SITUS is designed for environments where outputs must hold up years later—in audits, legal reviews, and lender scrutiny.
Glass-Box Accountability
Every claim traces back to:
- →source_id (origin document)
- →evidence_ids[] (photos/snippets)
- →timestamp + version
- →review_status (draft/staged/approved)
- →override_rationale (when humans override engine)
Provenance states: Verified (modern evidence + approval), Resurrected (from archive, unverified until confirmed), Derived (inferred, must show lineage).
Hierarchy of Truth
When sources conflict, SITUS enforces authority order:
- →1. Construction/architectural docs (as-builts, specs, permits)
- →2. Physical assessment evidence (photos, measurements)
- →3. County/municipal records (assessor, GIS, permits)
- →4. Third-party reports (previous PCAs, environmental)
- →5. Client-provided data (rent rolls, spreadsheets)
Higher authority wins by default. Override requires rationale + reviewer approval. Same-level conflict → flag for human. "Unknown" is always valid—never force a guess.
Human-in-the-Loop
Models suggest. System validates. Humans decide.
- →Every approval and override is logged with rationale
- →Permanent audit trail for defensibility
- →Override requires reviewer approval—never auto-applied
Nothing ships without human sign-off. LLMs propose—the system decides and proves. Every override logged with rationale for defensibility in legal reviews and lender scrutiny.
Archive Resurrection
Ingest legacy PDFs and photo logs to create structured history on Day 1.
- →Chunk by doc type (tables vs narrative vs bullet lists)
- →Preserve provenance (page/section where possible)
- →Defaults to Resurrected/Unverified until confirmed
Resurrected data NEVER auto-promotes. Human confirmation required. Contradicted by modern evidence → Superseded (never deleted).
Reports Are a Byproduct. Truth Is the Output.
SITUS doesn't just generate PDFs—it writes structured, provenance-backed truth that feeds your institutional memory.
SITUS
Generation Layer
Recon → Checklist Stage → Cost → Report
"Detect roof membrane failure on Building A."
"Map Building A roof to portfolio-wide membrane dependencies."
GRID
Memory Layer
Portfolio Graph • Evidence Timeline
ATLAS
Strategy Layer
Portfolio Rollups • CapEx Planning
"Forecast replacement costs for all Tier-1 membrane systems."
SITUS deliverables are instant. The asset intelligence compounds over time.
SITUS writes approved checklist evidence, staged investigations, and cost-linked outputs into a consistent record. GRID and ATLAS integrations are coming soon.
Powered by Approved Outputs
SITUS only generates reports from confirmed findings. Every claim traces back to source evidence and reviewer sign-off.
Reports are customized for each firm based on:
- ASTM E2018-24 — Baseline property condition assessment standard
- Industry standards — EULs, unit costs, and common practice
- Client scope of work — Client-specific requirements and preferences
ASTM E2018-24 also references olfactory and auditory observations. SITUS supports visual evidence; qualified professionals on site handle those elements.
Available as a standalone workspace or integrated into your existing ERP via API.
No unverified AI guesses
Every output derives from confirmed findings and reviewer sign-off—never from raw model output.
Eval harness: 100% claims map to inputs. Citation integrity. <5% drift across runs. Guardrail enforcement. Deploy threshold: 95% pass, 0 critical FAILs.
Satellite Investigation to Checklist in One Click
Step 1
Map investigation to checklist field
Step 2
One click adds it as a Pre-Site Action
Step 3
Defect-tagged rows are cost-ready
Autofill safety: current checklist value stays; AI suggestion appears below for reviewer decision.
Workflow
AI Research runs pre-site (Research tab). Drive-By AI runs on-site (Site tab). Staged findings flow through Findings before report output.
Governance
Additional capabilities
Deploy the
engine.
We’ll map your deliverables and stand up a workspace.