Urbyn + LenderGraph
2025 — 2026Source-backed data products
Fragmented public records and lender material turned into searchable products with traceable sources, review queues and honest publication boundaries.
The problem
Useful information about property and lending sits in public records, filings and lender material — technically available, practically unusable. It is fragmented across sources, inconsistently formatted, and stale in ways that are not visible from the record itself.
The hard constraint is not extraction. It is that a data product people rely on has to be able to show where each fact came from, and has to refuse to publish what it cannot stand behind.
The approach
Pipelines that collect and normalize from many sources while keeping provenance attached to every field, so any published value traces back to the document it came from.
A human review queue for the cases where judgment is required. Automation handles volume; a person handles ambiguity. The system is explicit about which is which.
Publication boundaries enforced in code — records that fail confidence or sourcing checks do not reach the product, rather than reaching it with a caveat nobody reads.
Where it landed
Searchable products built on records that can be traced back to source.
A review workflow that scales the judgment step instead of pretending it is unnecessary.
Next step
Tell me where the business is losing time.
No pitch deck and no discovery-call funnel. Describe the bottleneck and I will tell you plainly whether it is worth building something for, and roughly what that would take.