Institutional IC memo depth
The memo now spans 138 Phase 3 modules, including live market context, capstone verdicts, risk diagnostics, sector overlays, capital stack recommendations, and print-ready diligence sections.
Current scope | May 2026
Where Underwrite still draws the line.
Underwrite is now a broad deterministic underwriting layer: the engine includes 138 Phase 3 modules, live FRED macro context, rent-roll economics, tax overlays, capital-stack diagnostics, and sector-specific diligence screens. The point of this page is simple: tell users exactly where Underwrite stops.
Commercial posture
Paid beta access starts at $49 per month. No automatic charge. Ali reviews fit and scope before activation. This page explains the diligence boundary behind that paid workflow.
392/392
regression assertions
138
Phase 3 modules
Live FRED
Macro feed
Public
Engine source viewer
What changed
The old caveat list is no longer the product.
The engine now covers many of the items that were originally listed as future work. The list below separates what is now modeled from the boundaries that still deserve outside diligence.
Rent roll and leasing economics
Tenant rent rolls, escalators, free rent, TI/LC, rollover downtime, net effective rent, and tenant concentration are modeled when the deal carries commercial lease detail.
Capital stack and maturity risk
Senior debt, optional mezzanine debt, refinance events, rate sensitivity, debt sizing, lender risk, refi-rate risk, and macro-aware structure recommendations are now surfaced.
After-tax and partnership overlays
Depreciation, cost segregation, bonus depreciation, recapture, capital gains, NOL carry-forward, state tax, GP co-invest, sponsor fee drag, and waterfall economics are explicit.
Live macro and Phase 3 context
Treasury curve, SOFR, CPI, inflation expectations, macro regime, bank lending, CMBS spreads, REIT sector read-through, liquidity, climate, and mandate context are layered onto the deterministic engine.
Asset-class intelligence
Hotel, office, retail, industrial, multifamily, mixed-use, data center, life sciences, student housing, healthcare, SFR, workforce housing, and other sector screens now have targeted analytics.
Operating execution and cost recovery
Insurance, lease recovery, utility exposure, opex controllability, repair reserves, tax appeal posture, and property-management execution now sit in one operating-risk stack.
Entitlement and site-condition screens
Zoning, parking access, mixed-use conflict, environmental site condition, climate insurance, and redevelopment context are now surfaced as diligence screens with clear non-legal boundaries.
Boundary doctrine
Four levels, one audit trail.
The product is strongest when the page says exactly what kind of answer it is giving. Underwrite separates computed engine output from diligence screens, analyst judgment, and external workstreams.
Calculated
Engine output
The deterministic engine computes the metric directly from inputs, live macro context, or derived deal results. These numbers are regression-tested and repeatable.
Flagged
Diligence screen
The memo identifies risk direction, likely review intensity, and questions to ask. These modules help sequence diligence, but they do not replace outside reports.
Reviewed
Analyst judgment
The analyst decides whether the source document, lease language, local practice, or sponsor plan supports the assumption. Underwrite preserves the trail.
External
Specialist workstream
Counsel, environmental consultants, surveyors, lenders, accountants, and paid data vendors still control facts that cannot be inferred from underwriting inputs.
Modeled vs external
A diligence map, not a black box.
current as of May 2026
| Domain | Modeled now | Flagged for review | External control |
|---|---|---|---|
| Insurance | Premium load, deductible shock, carrier renewal risk, placement route, escrow posture, and tenant pass-through risk. | Coverage exclusions, named-peril exposure, lender exception risk, and sponsor-absorbed NOI drag. | Binding quotes, policy forms, carrier appetite, broker market feedback, and lender insurance requirements. |
| Operations | Opex variance, controllability, utility reimbursement, repair reserve shocks, vendor exposure, and property-management execution risk. | Where sponsor leakage could come from recoverability gaps, service complexity, weak cost controls, or underfunded repair obligations. | Executed vendor contracts, operating manuals, property-manager track record, service-level terms, local labor pricing, and asset-specific maintenance plans. |
| Leases | Rent roll economics, escalators, free rent, TI / LC, net effective rent, rollover exposure, and tenant concentration. | Expense recovery leakage, co-tenancy exposure, unusual reimbursement language, renewal rights, and termination options. | Executed lease abstracts, legal interpretation, estoppels, SNDA review, and tenant-specific credit analysis. |
| Entitlement | Zoning posture, density sensitivity, parking relief, nonconforming-use cues, adaptive reuse fit, and likely timing friction. | Where approval risk could change schedule, lender posture, or reserve sizing. | Permitted use confirmation, municipal approvals, title exceptions, code interpretation, and counsel opinions. |
| Environmental | Prior-use screen, Phase II probability, vapor risk, floodplain cues, brownfield reserve sufficiency, and lender escrow read. | Funding gaps, expected delay months, remediation overhang, and reserve coverage weaknesses. | Phase I / Phase II reports, remediation estimates, environmental counsel, agency correspondence, and lender environmental conditions. |
| Market data | Live FRED macro data plus seeded, hybrid, and derived overlays for comps, liquidity, climate, demographics, and sector demand. | When underwriting relies on proxy data instead of directly licensed market feeds. | Paid comp databases, broker surveys, proprietary rent rolls, appraisal support, and local market interviews. |
| Fund structure | Deal-level after-tax overlays, waterfall economics, sponsor fee drag, GP co-invest, promote yield, and LP suitability. | Where investor-specific tax, side-letter, or fund-level mechanics could change the interpretation. | Partnership tax modeling, audit treatment, REIT compliance, side-letter administration, and fund accounting. |
Still outside scope
Where analyst diligence still controls.
Legal, entitlement, and title review
- The engine does not confirm zoning, permitted use, certificates of occupancy, nonconforming-use continuity, easements, title exceptions, litigation, or municipal approvals.
- A new entitlement screen can flag likely diligence intensity, but counsel and local reports still control the answer.
- Environmental site condition analytics can triage prior-use history, Phase II probability, vapor, floodplain, and timing exposure, but reports, surveys, and lender legal conditions still control the answer.
Third-party market data coverage
- Live FRED data is real and marked with provenance. Many private-market overlays still use deterministic seeded proxies until direct data licenses are in place.
- Cap-rate comps, rent comps, transaction liquidity, REIT read-through, supply pipeline, demographics, climate, and foreign-capital modules are diligence screens, not paid-data substitutes.
- The page intentionally distinguishes live, hybrid, seeded, and derived signals so users know what is observed and what is modeled.
Construction and heavy redevelopment
- Lease-up, capex yield, construction-cost inflation, climate retrofit, and development sensitivity exist, but Underwrite is not a monthly draw model.
- Hard-cost line items, soft-cost timing, interest reserve, retainage, GMP contract mechanics, change orders, and month-by-month absorption remain outside scope.
- Ground-up development and major repositioning still need a sponsor budget, schedule, and construction lender model beside Underwrite.
Lease-document complexity
- Tenant rent rolls are modeled, but CAM recovery waterfalls, expense stops, percentage rent, co-tenancy cascades, renewal probabilities, expansion rights, and termination options are not fully contract-modeled.
- The AI parser extracts source quotes, but it does not replace abstracting leases or reviewing executed documents.
- Retail, office, and industrial deals with unusual reimbursement language should still be checked against the source lease files.
Fund, entity, and accounting structure
- The after-tax layer is useful for deal-level economics, but it is not a partnership tax model, audit model, REIT compliance model, or fund waterfall ledger.
- Entity-level debt, side letters, blocker structures, clawbacks, catchups across multiple assets, and investor-specific tax positions remain outside scope.
- Users should treat tax outputs as a deterministic overlay for underwriting, not accounting, tax, or legal advice.
Workflow and governance
- Saved deals remain local-browser workflow unless the user intentionally shares a link. Team roles, approval routing, comments, field-level audit logs, and permissioning are not yet enterprise workflow features.
- Compare view and portfolio rollups help with side-by-side analysis, but they do not replace an investment committee system of record.
- Confidential Mode disables AI surfaces, but user organizations still own document-handling, redaction, and data-governance policy.
AI boundaries
- AI can parse broker material, draft executive summary language, and provide qualitative review. It does not compute the math.
- Every deterministic metric comes from the engine, not from the model. AI-generated language should be reviewed by the analyst before circulation.
- Source quotes make extraction auditable, but unreadable PDFs, missing rent rolls, or ambiguous IM language still require human diligence.
How to use this page
Underwrite is a deterministic underwriting layer, not a claim to omniscience.
Use Underwrite when you need fast, reviewable, IC-grade deal math with transparent assumptions, live macro context, and a memo that can be defended section by section. It is especially strong for acquisition screening, lender read-through, sensitivity framing, and investment committee prep.
Use specialist workstreams alongside Underwrite when the answer depends on legal permissions, executed lease language, construction schedule, tax entity structure, or paid private-market data. The best version of the product is honest about that boundary.