RUDY ZOLEZZIDallas, TX · production AI
AGENT HQ · THE TEAM IS IN

I build production AI that moves the numbers.

I design, code and ship AI systems end to end at a medical-legal practice that runs on protected health information (PHI). Exam volume and revenue doubled year over year with no new hires, and the practice now saves about $240K a year ($210K in salary costs avoided and $30K+ in software cut).

The roles I fit best are AI Automation Lead, Senior Technical Product Manager (AI) and AI-enabled Director of Operations. I have Claude draft the architecture and specs and Claude Code write the code, and nothing ships until its tests pass.

Start with case file 01, or scroll down to meet the agents in their offices.
Photo of Rudy Zolezzi Pixel-art version of Rudy
Rodolfo "Rudy" ZolezziDirector of Operations at a medical-legal practice.DALLAS, TX · EMAIL · LINKEDIN · RESUME
Rudy · the humanI build the systems and sign off on everything. The agents in the offices below do the routine work.
ROOM 00 · RUDY'S DESK
Case files

Systems I designed, built and still run, told as one case moving through them from start to finish.

CASE FILE 01

The doctor's one-stop shop LIVE

From claimant registration to a finished report and every state form, in one system the doctors actually use.

Open the case file →
The offices

These agents are a stylized version of the bot team I'm building. Five work for the practice and one handles my own sites and admin. None of them touch claimant data.

Scout, Sentinel and Ledger find the work. Quill drafts it, Atlas checks it, and I approve it before anything goes out.Click an office to open its profile.
Daily standup

Each morning the agents meet for a short standup about the week's work. It plays when you scroll to it, and you can click a name tag to open that agent's profile.

IN STANDUP
Click a name tag to open that agent's profile.
What they're modeled on

The agents above are a stylized version of real systems. Here are two of them.

10,000+
physician reports consolidated into a queryable database in a 100% air-gapped environment, de-identified by local models with a second model checking the work
6 agents
in a live multi-agent "Chief of Staff" system, where an orchestrator hands out and checks work on referral and market research, content, compliance, regulatory watch and vendor spend
LIVE SITES I DESIGNED, BUILT AND RUN · each opens in a new tab