The Questions We're Answering
How long does it actually take?
From first proposal to Certificate of Occupancy, track the real timeline of publicly financed affordable housing.
Where does time accumulate?
Before financing, during approvals, in permitting, or in construction? See exactly where projects slow down.
Which delays are systemic?
Identify patterns across hundreds of projects to distinguish one-off issues from fixable bottlenecks.
Mission
Track projects end-to-end so we can bring the right government capacity, resources, capital, and political focus to the places that actually determine whether we deliver on our housing goals.
Vision
A city where housing goals aren't just promises—they're delivered, because the full pipeline is transparent, tracked, and actively managed like the critical infrastructure it is.
Why now
Mayor-elect Zohran Mamdani has committed to building 200,000 homes in 10 years. We believe he can do it.
This is the first moment where the city can pair political will with a step-change in execution capacity. Modern data tooling and AI can turn fragmented records into real-time operational dashboards, speed up compliance-heavy workflows, and help teams spot problems earlier.
The result: a housing pipeline that's not just funded, but actively managed to deliver.
What We Publish
Project-level data and timelines designed to be useful for delivery teams and transparent to the public.
How We Built It
The ingredients are public, but the story is fragmented.
The Challenge
No single dataset describes a project from early planning through financing and approvals to permits and completion. Different systems use different identifiers (BIN, BBL, project name), and projects change addresses midstream.
Our Pipeline
We stitch together structured records (housing project data, building records, permits, certificates of occupancy) with the unstructured predevelopment trail (environmental review, land use documents) to reconstruct end-to-end timelines.
Validation
Because it's easy to produce something that runs but isn't correct, we pair assembly with validation—checking that key timestamps exist and appear in coherent order—so the output tells a consistent story about a real building.