Navy vet. Grew up on the water on Cooks Bayou. Spent years running a real estate business. Now I build software and AI-powered tools, most days ten to twelve hours. Not just real estate tools: tools that use AI in the work itself. You can tell one to move a closing date, and it shows what will change before it does anything.
Hand is my chief of staff, a GrokBot assistant from SpaceXSI. It handles my email, sits in on the team's build huddles, and writes up what the team is working on.
Sprint report from the retc project manager: five production releases, zero rollbacks, the last four overnight Friday into Saturday. Includes how each model did.
Read the full entryAn open-source website your AI assistant builds and keeps up to date. Ten ready-made looks, one approval from you before anything goes live, free on Cloudflare or on your own server. The code is on GitHub at github.com/iXanadu/porchsite. This site runs on it.
porchsite.devMulti-line business phone on the web, Mac, iPhone, and iPad. An assistant can answer and screen calls, and the line records, transcribes, and summarizes them. Live in beta with one company on it.
Transaction coordination for real estate teams. Coordinators connect their own mailboxes and work every file from one place.
A community media platform, built by my agent teams in overnight sprints. Soft launch of one location, by invitation, starts the week of October 4.
Runs my agent fleet: launches agent teams and manages them while they work. Native apps for Mac, iPhone, and iPad.
The tool I built to turn a script into finished video. Used on my own projects for now.
Agent-friendly websites: what's real, what's hype, and what I'm building
My team read the standards, the proposals, the log studies and the X chatter. Here's what agents actually use, what they ignore, and what I'm putting into Porchsite.
#agents#porchsite#websites
How I Run AI Agents as an Engineering Team
Claude, Codex, and Grok, Cursor, each in its own harness on its own subscription, managed the way I managed human developers.
Auditioning a cheap model on a security boundary
We ran a DeepSeek seat for about nine hours as one of six agents hardening the security boundary of a multi-tenant Django + PostgreSQL product, alongside Claude and Codex seats.
All my writing (8) →
Back to top
I grew up on Cooks Bayou. It was rural: the nearest family with kids my age was a fifteen-minute horse ride away. We've been a fishing family for generations. Nights were for floundering. Days were for scalloping and oystering.
I spent eight years in the Navy, mostly in IT, where I learned my first languages: COBOL, Fortran, and some assembler. After the Navy I was a programmer at CBN for seven years, then an independent consultant until my client brought me onto their team. There I was named co-inventor on several software patents for cash office automation, and later moved into sales. After that I went independent again and ran a hosting company that reached more than 200 monthly clients. I sold cars for a year, then went back to development at ForRent.com, managing software engineers. In 2009 I left to sell real estate and built a small team.
About a year ago I started building software with AI. A rough stretch in 2025 made me stop pretending time is infinite, and my dad turns 90 this December. So I'm not doing this as a brand launch. I run a team of AI agents the way I used to run people, and the work that matters gets the hours.