ZoomProp
Founder & CTO2022–Present
I founded ZoomProp and lead product and engineering. It gives real estate teams an AI phone line that answers every property call, with an operator desk behind it where people run the work those calls create.
- A per-team AI voice line that answers and places calls, captures leads, showings, and maintenance requests, and hands off to a person when it should.
- An operator desk for calls, clients, deals, 1031 exchanges, prospecting, websites, and properties.
- Property search over nationwide data, plus a browser dialer for prospecting.
- Agentic infrastructure that takes work from a request to a checked, shipped change.
The work#
ZoomProp is two halves that share one record. The voice platform is a branded AI line for each team. It answers inbound calls, qualifies buyers, renters, and maintenance requests, summarizes every conversation, and turns spoken asks into tracked work. The operator desk is where people pick that work up: a "Needs you" overview, calls, clients, deals, 1031 exchanges, prospecting, the brokerage website, a shared workspace, and properties.
I started it, and I own product and engineering end to end.
What I built#
- Voice. The phone line runs on Vapi and Twilio, with LangChain and LangGraph in the backend. The line only reports success when the underlying action actually succeeded. A stub or a queued job is never presented as done.
- The desk. A Next.js app with a backend-for-frontend layer in front of a Python FastAPI service. Every surface goes through a shared service layer instead of writing its own queries.
- Two spines. People and their intents live in one place. Coordination lives in an append-only event log that feeds jobs, call slots, and inboxes. Nothing forks a second list of leads.
- Search and data. Property search on Elasticsearch over nationwide data, a small Fastify service for address matching, and Postgres on Google Cloud.
- Dialer. A browser dialer inside prospecting, so outreach and the call record live in the same place.
- The deploy loop. Agentic infrastructure around the codebase, where coding agents pick up work, run it through checks, and hand back a change ready to ship.
How I led#
I work across all four of my pillars here: the brand and the product, the architecture with my hands on it, the team and the delivery rhythm, and the agentic systems underneath.