Catalyze AI
The full platform behind a predictive seller-lead product: ingesting public records at scale, inferring the relatives likely to inherit a property, and scoring those properties for likelihood of sale.
Inferring an ownership transfer nobody has recorded yet
Inherited property is one of the strongest predictors of a near-term sale, but the signal arrives buried in public records that were never designed to be joined: death records, obituaries, and county property data, each with its own format, latency and spelling of the same human being.
The hard part is not storage, it is inference. Establishing that a decedent owned a specific parcel, and then identifying which living relatives are likely to inherit it, is an entity-resolution problem across noisy sources where a false positive means an agent contacts a grieving family about the wrong house.
A pipeline that resolves entities before it scores anything
We built the platform as a sequence of stages with clear contracts between them, so ingestion volatility never reached the scoring layer and the portal always read from settled data.
Normalising public death, obituary and property records into a common schema in PostgreSQL, with source lineage retained for auditability.
Matching decedents to owned parcels and inferring probable beneficiaries, tuned so precision is favoured over recall where a wrong match is costly.
Ranking resolved events by probability of sale, so the product surfaces the small set of leads worth acting on rather than everything it found.
A React portal that delivers scored, territory-filtered leads to real estate agents, with the account and delivery mechanics behind it.
The entire product, built and shipped by one team
Rainier delivered the platform end to end as prime: data pipeline, inference layer, database design and the customer-facing portal. The product is in production and sells to real estate agents on a subscription basis.
Have a system with the same constraints?
Send us the scope and the compliance requirements. We will tell you what the architecture should look like before you commit to anything.