Property data with provenance
Most property data vendors return a number. Ammasa returns a number and tells you where it came from, when it was collected, and what you are licensed to do with it.
That distinction matters the moment a customer disputes a value, a compliance team asks where a figure originated, or a downstream product needs to know whether a field can legally be shown to an end user. Provenance is not a reporting nicety — it is what makes the data usable in a regulated product.
Coverage
- Valuation — property value signals and assessment context.
- Flood & climate — FEMA zones and climate risk exposure.
- Permits — live building permit feeds from municipal systems, with explicit "no verified source" status where a jurisdiction does not publish one.
- Schools — nationwide school assignment and performance scoring.
- Crime — neighborhood incident data.
- Insurance — risk signals relevant to premium modeling.
Coverage is published transparently per county and per city rather than claimed in aggregate, so an integrator can see exactly what they will get for their market before writing a line of code.
How we built it
Ammasa is an adapter-per-source architecture: every county portal, state agency and open-data endpoint gets its own adapter with its own timeout budget, retry policy and failure semantics. Adapters run behind a queue, results are normalized into a parcel-keyed schema, and each field carries a provenance record through to the API response.
The operationally interesting part is failure handling. A source that times out must never be reported as "no data exists" — that is how data products quietly start lying. Ammasa distinguishes between no record found, source unavailable and no verified source for this jurisdiction, and surfaces the difference to the caller.
Ammasa powers BeforeYouBuy Homes and is available to other teams as a standalone API.