I founded ZapCRE after paid consulting and engineering work for CRE brokerages revealed a recurring problem: producing offering memorandums from scattered deal materials. I built memo.estate as the MVP, then developed it into ZapCRE.
What I built
I built a LangGraph agent system that turned commercial real estate source files into full offering memorandum decks in 20 to 30 minutes. The product brings document generation and an editable deck into the same workflow; the screenshot shows the assistant adding an amenity map to an existing OM.
Model testing and fine-tuning
I used H100 compute nodes through UT Dallas and the UT System to test supervised fine-tuning datasets and run fine-tuning experiments for ZapCRE.
Outcome
Zapcre reached enterprise customers and roughly $150K ARR within three months of launch. The product grew out of memo.estate, the initial MVP shown below.
Henry.ai, a YC-backed company, explored an acquihire of ZapCRE. The deal did not close.

