Two live, self-built platforms that model the commercial real estate workflows I know from the inside. Below is a screen-by-screen walkthrough of each — what every view shows and what it does.
An end-to-end platform that carries a commercial real estate loan from application through post-closing across eight connected stages. Upload the deal documents and it runs its own independent underwriting — then flags exactly where the paperwork and the numbers disagree, so the lender judges the deal on the right figures.

The command center for a live deal — Riverside Office Park, a $14.3M office acquisition. Loan amount, LTV, DSCR, open risk flags by severity and insurance status sit side by side, above a clickable pipeline that shows exactly where the deal stands, from Document Review through Post-Closing.

Drop in a loan agreement, rent roll, appraisal or operating statement — a single file or a full loan package — and the AI reads it and extracts the key terms. It handles PDF (including scanned), Word, Excel, CSV and images up to 50 MB, then feeds the findings into the rest of the deal.

Sizes the loan three ways at once — maximum LTV, minimum DSCR and minimum debt yield — and shows which constraint is binding. Here debt yield caps the loan at $16.5M, while stress tests warn that DSCR drops below 1.0x at +250 bps and breakeven occupancy runs above the comfort level.

A full cash-flow waterfall from gross potential rent down to cash flow after debt service, headlined by NOI, DSCR, LTV, debt yield and cash-on-cash — with a line-by-line income and expense breakdown priced per square foot.

Assembles a lender-ready credit memorandum — recommendation, an explained risk grade (here 3/10, strong · low risk), covenant compliance and a full deal summary. The narrative is editable and exports straight to PDF.

A conditions-to-fund tracker showing readiness to fund (90%, 18 of 20 conditions satisfied), insurance readiness, and every outstanding item — so nothing blocks the lender at the last minute before funding.
A loan-level escrow workspace built around real servicing logic. It walks a commercial loan's escrow (impound) account from closing setup, through tax reconciliation and the annual escrow analysis, to tracking every disbursement — with an AI analyst flagging shortages, variances and lapsing coverage along the way.

The account at a glance — escrow balance, monthly payment and annual obligations — with the AI surfacing issues like a tax-reassessment variance and a hazard-insurance renewal due in days, plus a 12-month projection of where the balance is heading.

Stands up the escrow account at closing: annual need, monthly payment, cushion target and initial deposit, alongside a checklist of the six documents required — tax bills, insurance binders, flood determination — each routed to the module the AI fills.

Reconciles the tax bill against what's escrowed, parcel by parcel and installment by installment. Here a county reassessment lifts the second installment by $2,800 — a shortage the tool carries straight into the borrower's payment as +$233/mo.

The annual analysis engine: it projects the account forward 12 months, finds the lowest balance and tests it against the lender's cushion to decide whether the payment holds, rises (shortage) or falls (surplus). Here the low point meets the cushion exactly — on target.

The pay-out side — every tax installment and insurance premium leaving escrow, with due dates, payees, amounts and status (paid, scheduled, due). Funds release about 15 days early so nothing lapses or draws a late penalty.