Every public record about a NYC building, in one place.
Weverit pulls every public record about a New York City building into one place. Type the address, get a report — violations, complaints, building data, neighborhood signals. Free, no signup.
Public, but not accessible.
NYC publishes more municipal data than any other American city. Every Housing Preservation and Development violation, every 311 service request, every Department of Buildings permit issued, every police complaint — all public, all free, all sitting on data.cityofnewyork.us.
But the data is scattered across dozens of databases, each with its own search UX, its own column naming, its own access pattern. A tenant evaluating a building before signing a lease has 24 hours and an iPhone, not a SQL prompt. The information that protects them is technically public and practically inaccessible.
Weverit closes that gap. We query the official sources, summarize the answers in plain English, and link every claim back to the dataset of record. The same data the city publishes; the same data a determined tenant could find in three hours of clicking. Just compiled, in fifteen seconds, from one address bar.
The lines, drawn clearly.
We do:
- Pull data from official NYC, state, and federal public records
- Show what's there in plain English, with the original numbers preserved
- Update reports daily
We don't:
- Show buildings for rent (we're not a listing service — try StreetEasy, Zillow, or Apartments.com)
- Rate buildings on a 5-star scale (the data speaks for itself; star ratings hide the texture)
- Sell your information (we don't collect it)
- Charge for the basic report
- Take broker money, advertising money, or referral fees from any third party
Official city and federal sources.
All our data is sourced from official NYC, state, and federal public records. We hold no data we didn't fetch from a public source.
Two products, one editorial standard.
For news articles:
- A draft is generated from a freshly-fetched data snapshot
- A human editor reviews, verifies every claim against the source, and edits before publication
- Each article includes a methodology paragraph
- Sources link back to the original dataset record
For building reports:
- Data is fetched live from the city's APIs at request time (or served from cache)
- An AI summary is generated to translate the numbers into a paragraph
- The underlying numbers are always shown alongside — the AI summary is the abstract, not the report
For the full editorial process, including how AI-assisted articles are produced and verified, see our methodology page.
Apasov & Sons LLC. New York City.
Independent. No broker affiliations. No advertising. No data resale.
Who's behind Weverit.
Aleksei Apasov is the founder of Weverit. He built the platform to compile New York City's public records — HPD violations, DOB filings, DHCR rent-stabilization history, 311 complaints, ACRIS ownership, and more — into clear, independent due-diligence reports for renters. Before Weverit, he founded and operated news and information websites, giving him a background in turning public data into useful, trustworthy products. Weverit is tenant-side, with no broker affiliations and no advertising — the goal is to give renters the same public-records picture a landlord or broker already has.
Where to write.
Where AI helps, and where it doesn't.
Some Weverit news articles are drafted with assistance from a large language model (Anthropic's Claude) and reviewed by a human editor before publication. We disclose this in each article that uses it.
We do not use AI to invent any data. Building reports are direct extracts from NYC's public databases. AI is only used for:
- The summary paragraph at the top of a building report (an abstract of the data, not new claims)
- The FAQ section of a building report (questions a tenant might ask, answered using the same data shown elsewhere on the page)
- Drafting news articles (always edited by a human, with every numerical claim linked to a public source)
Numerical drift is a known failure mode of AI text generation. We mitigate it with two layers: model temperature set to 0 for any factual extraction (deterministic outputs), and an explicit copy-exact-numbers rule in the system prompt. We do not run AI as the source of any number you read on the site.