A day-by-day record of what landed in Underwrite, kept by hand. Page 23 of 23.
Want to verify? The six files the decision depends on are rendered from the live source at /methodology/engine.
Filter by engine, UI, AI, ops, marketing or fixes.
Monte Carlo, Deal Strength composite score, sensitivity curves, break-even cushions, risk-adjusted returns, tenant concentration, hotel ops overlay, GP coinvest sweep, debt sizing optimizer, NOI decomp, a below-NOI leasing-capex parser, and 15 more analytics. The day the memo went from 'pro forma + waterfall' to 'institutional IC packet.'
Six dynamic-route pages (Hotel, Multifamily, Office, Retail, Industrial, Mixed-Use), each with the asset's native operating metrics, property-type-aware validation bands, the matching stress scenarios, a computed example deal, and class-specific Phase 1 simplifications.
Exhaustive list of every institutional CRE feature deliberately not modeled in Phase 1, by asset class. The trust play: 'we'd rather tell you exactly what we don't do than pretend we do everything.'
Hotel Demand Shock, Office WFH Shift, Industrial Cap Expansion, Insurance/Tax Spike (multifamily/mixed-use), Anchor Tenant Loss (retail). ScenarioBar filters by property type so users see only stresses meaningful for their deal.
Workbook gains OPERATING_SNAPSHOT, VERDICT_BREAKDOWN, LENDER_VIEW, and TORNADO sheets matching the IC memo institutional surfaces. Every sheet has 'Export generated (UTC)' timestamp. Filename appends YYYY-MM-DD suffix. RETURNS sheet includes after-tax decomposition. Workbook metadata sets the Title to the deal name for clean Explorer display.
Shared SiteNav + SiteFooter across every public page (tool, methodology, examples, beyond-phase-1, engine, asset subpages, changelog, compare). Active-state gold underline indicates current page. Footer email and getunderwrite.com are now real hyperlinks.
Parser maps source language by asset class: 'keys'/'rooms' → residential_units for Hotel, ADR + occupancy → monthly per key, NNN passthroughs flagged in notes for Industrial, etc. Reduces 'one shoe fits all' on the AI extraction layer.
Methodology, examples, source-quote audit trail
Thirteen-section walkthrough of every formula the engine uses to compute the IC memo. Cherry Street worked example end-to-end. Table of contents with deep-link anchors.
Each AI-extracted value pairs with a verbatim quote from the source text, server-validated to be a literal substring. Click any quote to highlight the span in the textarea. Direct answer to 'generic data back, unauditable.'
Splits Project IRR into operating yield (avg NOI / cost), asset appreciation (annualized sale-value growth), and leverage premium (equity IRR minus unlevered IRR). Surfaces WHERE the return comes from across tool + memo + compare + Excel.
Eight institutional threshold tests with pass/miss status pills: Project IRR ≥ 15%, Min DSCR ≥ 1.0x (critical) plus institutional 1.2x, LTV ≤ 75%, Yield on Cost ≥ 6%, positive leverage Y1, exit ≥ entry, hold ≤ term (advisory). Critical highlighted; verdict is exactly the conjunction.
Debt Yield (Y1), Break-Even Occupancy, Occupancy Cushion, Refinance Test at exit. Surfaces the lender's view of the deal, distinct from equity returns. Color-coded passes / misses.
Hotels show ADR / Occupancy / RevPAR / Cost per Key. Industrial / Office / Retail show rent and cost per SF. Multifamily shows rent and cost per unit. Mixed-Use shows hybrid. Tool + memo + compare all use the same shared component.
7×7 IRR matrix across exit cap × rent growth axes embedded in the memo with cream-to-gold heat ramp. Reveals joint sensitivity that a single-input tornado can't.
Stacked leverage risk (high LTV + thin DSCR + cap compression), refinance dependency, double-growth bet, property-type-aware property management bands. Surfaces compound risk patterns instead of just single-input flags.
Disables AI parser and AI deal review for sensitive IMs. Tool stays fully usable on manual inputs; every metric still computes. Trust panel surfaces the active state in the audit column.