I run a 2,600-unit leasing operation. I build the AI tools my team uses to run it.
Twenty years in luxury multifamily leasing in New York and San Francisco, the last eight managing teams of 8 to 15. I taught myself to build software, and now I ship the tools instead of writing requirements for someone else. I build with Claude Code, Claude Cowork, ChatGPT, and Codex, and I know what it takes to get a non-technical team to adopt a new tool.
Leasing Deskfront deskTour queueFollow-up pipeline
Shared live database
Tour CompanioniPad
Published showsheet both apps
Tours started on iPad Desk queue
Finished tours follow-up pipeline
Results
- 100 tours a week Leasing system
- 8 of 8 staff use it daily Leasing system
- ~70% to 100% of tours followed up Leasing system (before: busy periods)
- +28% revenue, year over year TRS, Amazon brand
- 15 of 15 pilot customers rated it “very useful” TRS Resource Centers
Selected work
-
~70% Before, busy periodsof tours followed up100% Nowplus a second follow-up
No missed leads after a tour. Leasing Operations System
Five Class A buildings, Long Island City, NYC
Two connected apps on one live database, used daily by all 8 team members across 100 tours a week.
-
+28% revenue year over year. TRS: one reconciliation tool to an AI-powered operations stack
Family-owned brand selling SDS and estate-planning binders on Amazon
A two-year engagement: ~3 days a month of manual work removed and $2–3K/month in ad spend saved.
-
3 failure modes found and fixed. Job Search Agent in Claude Cowork
For my own job search
A scheduled agent across three ATS APIs and Gmail, and the debugging that fixed its invented job titles.
- More builds: five more, including a diagnostic engine for Amazon CPC ads and a rental-comps pipeline.
- How I build with AI: plan in one model, build in another, review in a third.