Skeptical property due diligence
The listing says two units. The assessor says one.
Paste a listing URL or an address. We read the public record and the listing itself, underwrite the property at rents the neighbourhood actually pays, and hand back a verdict, the red flags behind it, the regulation that binds it, an offer range — and an explicit list of what we could not verify.
Free plan includes 1 report a month. No card to start. The sample below is 72 Sunset Ave, Newark NJ 07106, asking $699,500 — a real listing, not a mock-up.
The actual output
This is the report, not a picture of one.
Every element below is rendered by the same components the product uses, from the same engines, on 72 Sunset Ave, Newark NJ 07106 — a real listing asking $699,500. Nothing here is a mock-up, and nothing here can say something the report would not.
One of the 26 findings, expanded exactly as the report expands it — the explanation, what would resolve it, the evidence the rule fired on, and the records behind that evidence.
Deal-breakers
1Walk, or make the whole offer conditional on resolving it.The listing markets this as a 2-unit property. The assessor records 1 — the building description reads "2SF3UG", which decodes as two-story, single family, 3-car unattached garage. If the additional unit is not legal, the income case collapses: you would be underwriting one unit's rent against the whole purchase price, and a lender will not fund a unit that does not legally exist. Assessor descriptions do go stale, so this is strongly suggestive rather than conclusive — but the burden of proof sits with the seller, and the answer changes the value by hundreds of thousands of dollars.
What to verify
- The certificate of occupancy establishing the legal unit count — the document, not the agent's word.
- The zoning and permit history from the municipal construction-code office.
- A public-records request for all permits, inspections and certificates on the parcel.
- The assessor's classification of neighbouring parcels on the same block, to check how the code is used locally.
Evidence
- classification.assessorUnits
- 1
- classification.listingClaimedUnits
- 2
- assessor.useCode
- 2SF3UG
- classification.legalUnitsBestEstimate
- 1
The other 25 findings, the transaction history, the rent reality check, the scenario matrix, Newark’s rent control, the tax intelligence and all 10 items this report could not verify are in the full sample report, with 17 sources and no sign-up.
How it works
Four steps, and the two that take time say so.
There is no data feed behind this and no licensed MLS access. It reads the same public records a careful buyer would read, in the same order, and then does the arithmetic properly.
Paste a listing URL or an address
Either works. A listing URL is preferred because it gives the pipeline the seller’s own claims to check the record against — the square footage, the unit count, the rent projection, the remodel year.
The pipeline reads the public record and the listing
The address resolves to a parcel; the assessor record, the recorded sale history and the construction-permit feed are pulled from their own sources; the municipal regulation entries are read from the knowledge base; the listing page is fetched and parsed.
This takes minutes, not seconds — roughly two to six. Records are fetched one at a time and politely, and the job reports its progress while it runs. Nothing is instant, and anything claiming to be is not reading a county record.
Deterministic engines underwrite it
The rent basis, the operating expenses, the scenario matrix, the coverage ratios, the price ceilings and every red-flag rule are ordinary tested code. The language model’s only job is turning unstructured pages into typed facts.
The model never does the arithmetic. Given the same facts, the report is the same bytes every time — which is what makes a checked-in sample report possible at all.
You get a verdict, the flags, an offer range and a checklist
Buy, Negotiate to a range, Walk, or Insufficient data — with the rule that produced it recorded, the findings that decided it, and a recommended offer with the conditions to attach to it.
And a list of what could not be verified. The sample report carries 10 such items, each with why it matters and how to check it — a phone number for the construction-code office, not a shrug. They become a checklist you work through.
What makes it different
A calculator multiplies your inputs. A listing site republishes the seller’s.
Neither one goes and checks. The arithmetic is the easy half and it is not where deals go wrong — deals go wrong on a unit that does not legally exist, a renovation nobody permitted, a rent projection a quarter above what the street pays, and a tax bill that moves the day the deed records.
Assessor versus listing
Unit count, square footage, use code, year built and improvement value are cross-checked against what the listing markets. Where they disagree, the disagreement is the finding — and the burden of proof sits with the seller.
Listing says 2 units; the assessor records 1. The income case depends on which is right.
Transaction forensics
The recorded sale history is read for signatures rather than printed as a table: flip velocity, prices that sit outside the trades around them, long escrows closing far under ask, relistings that reset days-on-market, the same property listed twice under different agents.
Bought at $500,000 and relisted 11 days later, up 39.9%. Little or no work is possible in eleven days.
Municipal regulation, not state summaries
Rent control and whether it binds at this unit count, vacancy decontrol, just cause, rental registration and inspection, lead paint. Read municipality first, then county, then state — with the citation and the date it was last checked against the primary source.
Rent control caps increases at 4%. Rental registration and inspection regime applies. Both change what the property earns.
Tax intelligence
The assessment against its immediate neighbours, the county equalization ratio, revaluation risk, and — where the jurisdiction reassesses on transfer — what the bill becomes at the price actually paid, applied to every row of the scenario matrix rather than to the asking price alone.
Assessed at $277,100 against a $699,500 ask, and at 1.88× the neighbouring parcels.
What it could not verify
Every report ends with the things that could not be established from a screen, ranked by how much they matter, each with why it matters, how to check it, and who to call. An estimate is labelled an estimate wherever it appears.
The legal unit count, and the certificate of occupancy that would establish it.
The list of 10 unresolved items is the part most products leave out, and it is the part a buyer actually acts on. A report with nothing on that list would be a report that has not understood the question.
Coverage
Full public-records coverage in New Jersey and New York City. Nowhere else, yet.
Every row below is read from the market adapter itself when this page is built, so it cannot promise a source that is not wired up.
| What gets checked | New Jersey | New York City |
|---|---|---|
| Parcel identityResolving the address to the block and lot the public record files it under, so every later lookup is about this building and not a neighbour. | Read | Read |
| Assessor recordThe assessed value, the building description and the use code — the record that disagreed with the listing about the unit count. | Read | Read |
| Sale historyRecorded transactions, which is what makes flip velocity, price outliers and long-escrow discounts detectable at all. | Read | Read |
| Permits and certificatesThe construction-permit feed, including whether a certificate of occupancy has ever been issued. | Read | Read |
| Tax intelligenceEqualization ratios, effective rates and whether the assessment follows a sale — what the bill becomes at the price you pay. | Read | Read |
| Municipal regulationRent control and whether it binds at this unit count, just cause, rental registration and inspection, lead paint. | Read | Read |
| Rent comparables from a public recordSigned leases from a public source. No market has this: comparables everywhere come from national listing portals, which publish asking rents. | Not yet | Not yet |
New Jersey
New Jersey (NJParcels assessor + DCA permit database)
- Rent comparables are gathered from national listing portals, not from a New Jersey public record, and are estimates rather than leases.
- Equalization ratios and effective tax rates are transcribed from the NJ Treasury county tables and carry a citation rather than a live retrieval.
New York City
New York City (GeoSearch + PLUTO + DOF assessment roll + ACRIS + DOB permits)
- New York City does not reassess a property when it is sold, so the tax figures in this report are what the seller pays now rather than a projection off the purchase price. What can move the charge is the assessment phase-in cap, not the sale.
- The annual property tax figure is arithmetic over two Department of Finance publications — its billable assessed value and its published class rate — because New York issues no per-parcel statement as open data, and the class rate available may be a fiscal year or two older than the assessment roll. Confirm the figure against the seller's own statement before underwriting on it.
- Rent comparables are gathered from national listing portals, not from a New York public record, and are asking rents rather than leases.
- New York required no occupancy certificate for buildings erected before 1938, and the city's published file of them only opens in 2012, so for an older building this report draws no conclusion from the absence of one either way. Order it from the Department of Buildings for the parcel if how many homes the building legally contains matters to the deal.
A property outside both markets still gets a report. A thinner one, labelled as such.
There is no adapter, so there is no parcel lookup, no assessor record, no recorded sale history and no permit feed — which means the assessor-versus-listing check, the transaction forensics and the certificate-of-occupancy question simply cannot run. What remains is generic web research over whatever the listing and the open web publish, extracted into the same typed facts and underwritten by the same engines, with the local regulation entries applied where the knowledge base has the jurisdiction.
Every check that could not run is named in the report’s “what could not be verified” list, because “no permits found” and “this market has no permit source” are different sentences and only one of them is a red flag. If that is not enough to underwrite on, the report will say so rather than fill the gap with an average.
Pricing
One free report a month. Unlimited for the price of an hour of anybody’s time.
A report costs real money to produce — an agent reading records, fetches against half a dozen sources — so the free tier is one a month rather than a trial that expires. The public sample reports need no account at all.
Free
$0 / month
One report a month, and every public sample.
- 1 report per month
- Full verdict, red flags and underwriting on the web
- Interactive scenario sliders (results are not saved)
- Checklists with your own notes
- Watermarked PDF export
- Sample reports
Pro
Most investors$79 / month
Unlimited due diligence for one investor.
- Unlimited reports
- Side-by-side compare
- Saved scenarios per property
- PDF export without the watermark
- Alerts (as they ship)
Team
No price yet
Seats, white-label PDFs and client share links. Not yet available.
- Everything in Pro
- Multiple seats
- White-label PDF export
- Client share links
- API access
Not yet available. Pro covers a single investor without limits.
Free and Pro are the two tiers that exist today. Team is listed because the entitlements are built and the seats are not — there is no price and nothing to buy, and saying so is cheaper than an enquiry form that goes nowhere.
Questions
The five worth asking before you pay.
Is this investment advice?
No, and the report says so on every copy: “Market research and financial modelling, not investment advice. Every figure here needs verifying against primary documents before you commit — particularly the certificate of occupancy, the rent control status and the current tax bill.”
What it is: market research and financial modelling. It gathers facts from public records and the listing, marks each one with its source and the date it was retrieved or labels it unverified, runs deterministic arithmetic over them, and applies a published rule set whose verdict logic is written down and whose every call records the rule that produced it. It is not registered to give advice, it does not know your tax position, and it never tells you what to do with your money — it tells you what the record says and where the record and the listing disagree.
Where does the data come from?
Public records and the listing itself. In New Jersey: the NJParcels assessor record for the block and lot, the statewide construction-permit database that pulls from the municipality’s own system, the NJ Treasury county equalization tables, and the curated municipal regulation entries. In New York City: the Department of City Planning’s address directory and PLUTO for the parcel and its unit count, the Department of Finance assessment roll for the tax class — the difference between a three-family and a four-family is the most expensive thing about a New York purchase — ACRIS for the recorded deeds, the Buildings Department’s permit and occupancy-certificate files, and Housing Preservation and Development’s multiple dwelling register. Rent comparables come from national listing portals in both markets, which publish asking rents rather than signed leases, and are labelled as estimates wherever they appear.
There is no licensed MLS feed and no purchased data set. Every fact in a report carries the URL it came from and the date it was fetched, or it is on the unverified list — the sample report cites 17 sources. Two operating lines are estimates by construction, because nobody publishes an insurance premium or a water bill per address; both are marked “estimated” and both step aside the moment you enter a real quote.
What happens when a listing portal blocks you?
It is recorded and the report tells you. The portals refuse automated requests routinely, so the listing is resolved in layers: the URL you pasted, then other portals carrying the same property, then listing text you paste in yourself. The facts record which layer answered — a page was read, every portal refused, or nothing was fetched and you supplied the text — and the report never implies a page was read when none was.
What does not happen: no headless browser, no rotated user agents, no proxy pool. A portal’s refusal is answered with another portal, your own copy of the listing, or an honest gap. The public-records half of the report — assessor, deeds, permits, tax, regulation — is unaffected either way, and in New Jersey that is where the deal-breakers usually are.
How current is the regulation data?
Every rule carries the date it was last checked against its primary source, and that date is printed beside it — in the report and on the public regulation pages. Anything not re-verified within 180 days is marked stale on sight; staleness is computed when you read the page, so nothing goes quietly out of date because a job did not run.
Entries are curated and human-reviewed before publishing, and they live as checked-in records as well as in the database, so an unseeded deployment cannot produce a report with no rent-control section for a rent-controlled city. Ordinances still change faster than any database: the citation is there so you can open the ordinance yourself, and the report asks you to.
Can I cancel?
Any time, from the billing page, in two clicks. Cancelling means the subscription does not renew — you keep Pro until the end of the period you have already paid for, and then the account becomes Free again. Nothing is deleted: your properties, your reports and your checklists stay where they are and stay readable, you simply go back to one report a month, and every public sample.
If a card fails, the account keeps its plan for 7 days from the failure while the payment is retried, rather than locking you out on the morning of a closing.
The difference between $699,500 and $490,000 is one afternoon of reading records.
That is one property. Run yours before you write the offer, not after the inspection.
1 report a month on the free plan. No card, and the sample reports need no account at all.