Real estate · 2026 · Self-built prototype

A permit-velocity index that names its own sample size

A prototype that normalizes plan-review times across many jurisdictions into one comparable median an estimator or analyst can put in a fee scope, with every number carrying its sample size, its confidence, and its method.

Sector: Real estate
Year: 2026
Engagement: Self-built prototype
Team: Solo build, with Claude Code

Preconstruction estimators and pipeline analysts are all asking the same question in different words: how long will this jurisdiction take to issue, and how does that compare to the one next door. The answer is usually a guess dressed as experience. We built a proof that turns it into a number with a method behind it.

What we built

An index that ingests permit records across jurisdictions and returns the median days from application to issuance for a jurisdiction, permit type, and value bracket, normalized so the numbers are actually comparable. It runs as a queryable service and as an MCP tool, so a person and an agent can both ask "what is the plan-review spread between these two counties" and get the same cited answer.

What we found

The differentiator is transparency, not coverage. Anyone can publish a median; almost nobody publishes the sample size and confidence behind it. A median over eleven permits is not a median over eleven hundred, and pretending otherwise is how analysts get burned.

The boundary, stated as a feature

One jurisdiction is permanently excluded from coverage: the one where the founder holds a public-sector GIS role. That exclusion stands while the day job exists, and we state it in the product rather than hope nobody asks. It is a cleaner conscience and, for a client, a signal about how we handle a conflict when we have one.

What we learned

Permit data quality varies wildly. Some jurisdictions publish clean open data; some publish scanned PDFs. Where the data is too thin to trust, the honest output is "not enough data," and the index is built to say that instead of inventing a confident number.