The week goes to inputs
Transcribing a rent roll, reconciling an OM against a statement that disagrees with it, rebuilding a model because the unit count was wrong. The judgment you are paid for happens in the last hour.
Underwrite a commercial deal in an afternoon instead of a week — multifamily, office, retail, industrial or ground-up — with every number traceable back to the page it came from.
Not by how many deals you see — by how many you can actually get through. Every deal that eats two days is three you never opened, and the one worth doing this quarter might be in the pile you did not reach.
Transcribing a rent roll, reconciling an OM against a statement that disagrees with it, rebuilding a model because the unit count was wrong. The judgment you are paid for happens in the last hour.
The broker sent it to a list. Whoever comes back first with a credible number gets the conversation.
An investor, a lender, your own committee. Weeks later, in a spreadsheet of typed figures, "where did this come from" is a hard question.
A rent roll in XLSX, a T-12 as a scanned PDF, an offering memorandum, a P&L, seller disclosures. Cygni reads them and attaches to every figure the page or the spreadsheet cell it came from.
Document data extractionUnit mix, rent roll, revenues and expenses — each shown beside its source, each totalled from the document itself. Where the offering memorandum and the rent roll disagree, you pick which one the deal is built on, seeing what each actually says.
How confirmation worksRent growth, vacancy, expense inflation, property tax reassessment, exit cap. You review them before anything is projected, then they run across your hold period.
Proforma analysisDebt sized by LTV or DSCR across as many tranches as the deal has, then distributions period by period through the preferred return, the capital return and the promote.
Investor returnsUpside, downside, value-add, refinance, or solved backwards from a return target — all on the same approved figures, so the difference between two cases is only the assumption you moved.
Scenario builderWorth being blunt about, because the category is not. AI is very good at reading a scanned operating statement and recognising that the third column is a trailing twelve. It is the wrong tool for compounding rent growth over ten years or working out an IRR — not because it usually gets those wrong, but because "usually" is not a standard you can underwrite against.
So Cygni uses it for what it is good at. It reads the documents and suggests the assumptions; you review and change them; and the money maths is calculated the way a spreadsheet calculates — the same inputs giving the same answer, every time, on every deal.
What changes between asset classes is which documents exist and which assumptions matter most — the workflow is the same either way.
Unit-level rent rolls, unit mix, loss to lease, turn schedules and renovation premiums.
Lease-by-lease rollover, TI and leasing commissions, downtime between tenancies.
Anchor and inline tenancy, percentage rent, recoveries and CAM.
Long leases with fixed escalations, and the exit assumptions that dominate a flat cash flow.
Distinct revenue streams under one asset, each trended on its own driver.
No operating documents to read, so the basis comes from research and market rent evidence instead.
A syndicator presenting to investors, a broker producing a summary for a client, a lender sizing against a projection, an analyst whose model will be picked apart in committee — the common requirement is not speed, it is being able to show where a figure came from.
The point is not that any single deal gets underwritten faster. It is that screening stops being the bottleneck — so the deals you pass on are ones you actually looked at, and the one you bid on comes with numbers you can walk an investment committee through, line by line, back to the page each came from.
It is the software an investor, syndicator or lender uses to turn a deal's documents into a defensible view of its returns: reading the rent roll and operating statements, projecting income and expenses across the hold, layering on debt, and distributing what is left through the equity structure. The difference between platforms is mostly where the numbers come from and whether you can trace them.
Multifamily, office, retail, industrial, mixed-use, and land or ground-up development where there are no operating documents yet. The way it works is the same for all of them; what changes is which documents exist, which drivers matter, and how the revenue is structured.
Three things a workbook cannot do. Every figure keeps a link back to the page it was read from, so you can still check it a month later. The formulas cannot be typed over by accident. And a downside case is the same deal with one assumption changed, rather than a copy of the file that quietly drifts from the original. What Excel does well — showing you everything at once — is why you can still export the result and work with it.
Yes. For land, ground-up development, or a deal where the seller has not sent anything yet, the basis comes from market research and rent comps instead of extraction — clearly labelled as the reference it is, rather than presented as a fact from a document that does not exist.
The fastest way to judge this is on a property whose numbers you already know by heart. Upload its documents and see what comes back.