Project Lab / Physical-world decision intelligence
Before design hardens. Before the deal closes. Before money gets trapped.
Stress-test the project before the project becomes expensive.
ExpenseIntel tests a place and a plan together. It looks for utility uncertainty, fragile assumptions, approval bottlenecks, missing evidence, hard-to-reverse commitments and real-world failure patterns—then tells you exactly what to verify next.
Verified locationPlain-English intakeEvidence depthSame-plan site comparisonProject watch
Decision lens
Builder lens: sequence the work before the work sequences you.
The underlying model stays the same. The lens changes which decisions, evidence and actions are emphasized.
Smart project intake
Describe the project like you would describe it to a person.
ExpenseIntel translates the narrative into editable project signals—use, area, timing, utility needs, risk triggers and planned interventions. Nothing detected from the narrative becomes a verified fact; you confirm the controls before the model runs.
Optional. You can still use the structured controls below.
00 / Project fingerprint
Describe the place and the plan once.
Every system below uses the same fingerprint. Address data is verified where possible. Utility capacity is never assumed: if you do not provide known service data, ExpenseIntel leaves capacity unresolved instead of fabricating an answer.
PROJECT—
FRAGILITY—
BOTTLENECK—
UTILITY UNKNOWN—
LOCK-IN—
Current decision framing BUILDER
Sequence approvals, utilities, scope and irreversible releases before construction hardens.
Switch Builder / Buyer above at any time. The core project scoring remains the same; actions and evidence priorities change.
01 / Utility Headroom
Can the site support what the project wants?
ExpenseIntel separates modeled demand pressure from known capacity. A utility with no verified service rating stays unresolved. That distinction is the point.
Electric demand proxy—Modeled from proposed use and floor area.
Known service headroom—Enter service amps to calculate screening headroom.
READ THIS CORRECTLYBubble size = modeled demand pressure, not utility capacity. “Headroom” appears only where the user supplied a capacity input. Engineering and utility confirmation remain required.
02 / Project Fragility
How many things have to go right?
—/100Run project
Fragility is a screening measure of dependency, uncertainty, schedule pressure and capital lock-in.
Single point of failure—
—
03 / Approval Bottleneck Radar
Where does friction concentrate?
Dominant bottleneck—
Radar scores identify where verification effort should concentrate; they do not predict municipal approval time.
04 / Undo Cost
Which decisions destroy optionality?
The expensive mistake is not always spending too much. Sometimes it is committing to a change that is costly to reverse before the project has cleared its real constraints.
CAPITAL LOCK-IN ↑
EASIER TO REVERSE → HARDER TO REVERSE
EXPENSIVE
but flexible
but flexible
LOCKED
capital
capital
LOW-REGRET
moves
moves
STICKY
choices
choices
Highest lock-in choice—Based on selected interventions.
Selected high-lock moves—Reversibility score ≥ 75.
Undo Cost is a normalized reversibility screen, not a construction-price estimate. Its job is to protect sequencing: clear uncertain gates before hard-to-reverse spend.
05 / Evidence Depth
How much of this answer is actually proven?
A strong score with weak evidence is still a weak decision. Evidence Depth separates verified/project-specific proof from public signals, user-supplied facts and screening assumptions—and shows exactly what would strengthen the record.
0Evidence depth E
The project is still assumption-heavy.
Mark an item only when you have a real document, written agency/utility response, inspection, plan review, quote or responsibility record behind it. The checklist is intentionally stricter than “someone told me.”
VERIFIEDADDRESS OPENProperty identity and parcel-level evidence.
PUBLICPUBLIC SIGNAL LIMITEDReturned authoritative/public data.
USER / DOCUMENT0 diligence itemsFacts and documents supplied by the project team or buyer.
MODELEDScreening modelTransparent inference that still requires project-specific confirmation.
Best next evidence to raise depth
06 / Failure Library
Which failure patterns does this project resemble?
Pattern matching remains a screen—not a probability. The new source layer begins connecting those patterns to public primary-source project records involving denial, expired entitlements, delayed construction and approval-path friction.
Pattern relevanceProject-specific match, not historical probability.
Next moveEach warning now ends with an explicit verification action.
Primary-source evidence layer
Real records behind the pattern library.
Loading official public records. The seed library is intentionally small and source-first; it can grow into a national corpus without pretending a handful of cases are statistics.
08 / Action Orders
Do not end with “verify this.” Tell me exactly how.
Every major risk should terminate in an executable move: who owns the answer, what to ask, and what document or response actually closes the issue.
Current lens: Builder action orders
09 / Same-plan Site Compare
Does the exact same project survive better somewhere else?
Hold proposed use, area, timeline, contingency, triggers and interventions constant. Change the site and compare fragility, unresolved utility evidence, top bottleneck, reversibility and modeled operating cost—not price alone.
Site A / current project
—
VSsame plan held constant
Site B / candidate
—
Fragility— / —
Utility unknown— / —
Top bottleneck— / —
Lock-in— / —
Modeled annual cost— / —
Current screen—
10 / Save + Watch
Turn the project from a one-time score into a living record.
Project Watch stores the project and scored snapshots in this browser. Re-run after adding evidence—or when returned public signals change—and save again to see what actually moved.
MODEL DISCIPLINEObserved facts stay separate from screening assumptions.
ExpenseIntel / Project Intelligence
A project should get harder to misunderstand every time new evidence arrives.Describe → verify → stress → compare → act → save → re-score