36.58° N, 86.52° W · Portland, Tennessee
I put AI to work on real business problems.I build secure, governable AI-driven processes for production systems.
I'm Joe. Nearly 20 years keeping production systems running and automating the work around them. Now AI runs projects on them, fast, inside controls I built. I do the same for a few local businesses.
Nearly 20 years in production systems, Navy ship to Azure SaaS platform where I own uptime. I wrote the rules for AI on infrastructure, then let it run the projects. I'm happy here, and the door is open.
- Bowling Green
- Franklin
- Portland
- Westmoreland
- Cross Plains
- Springfield
- White House
- Gallatin
- Goodlettsville
- Hendersonville
Who I am
An operator, first.
I started in the Navy: classified messaging for the Atlantic Fleet, then the network on a ship forward-deployed in Japan. Communications either worked or the ship was deaf. Uptime is the product.
Then about 20,000 people supported at Roche. Now a decade on a SaaS platform, on-prem and Azure, where I automate routine work and own uptime.
In 2026 the job changed. I stopped using AI on my projects and set it up to run them, with me approving. I live in Portland, on the Kentucky line.
Three kinds of work
-
Infrastructure
Keeping production systems up
Automated patching you can prove, backups you've actually restored, monitoring that pages the right person.
Career and case files -
AI enablement
Setting AI up to do the work, safely
The fence comes before the keys: its own identity, least access, a lane it can't leave, a person approving what matters. That's what lets me hand it whole projects.
How I work -
Local
Automation for businesses around here
A few clients at a time, Bowling Green to Hendersonville. We find the work that eats your evenings and automate it. AI reads and drafts; your people approve what goes out.
Find your kind of business
Real numbers
Real figures from my own work. Employer details left out on purpose.
- 6h 14m Disaster-recovery drill, failover to a verified application First complete drill in years, against a 24-hour target. AI ran it; I approved.
- 129 Changes merged in 26 days to rebuild patching 27 old scheduled tasks retired. AI did the work; I approved it.
- 31 min To pull a year of compliance evidence Auditor evidence, machine by machine. It used to be a project.
- 7 sec Cloud database relink Instead of an estimated 16 hours, after a byte-for-byte check.
- ~90 sec For leadership to approve an AI agent acting on infrastructure Once they saw the controls. Fences, not reassurance.
- 8% to 85% Navy advancement rate after I rebuilt the training 3,000 users, a team of eight. Systems beat heroics, even then.
It's not about whether or not it's going to make mistakes. Of course it is. The point is keeping human intelligence at the right place.
Side business
Morning Stack, a company I run with agents
It started as a pipeline to help my wife land her first American job. Now it's a subscription: three tailored application packages in each customer's inbox at 7 AM. Nothing auto-applies.
AI agents run most of the business; I review their work. Proof I can ship and run an AI product alone, next to a day job.
You don't have time to fix it. I do.
Quotes at the kitchen table, callbacks nobody made, invoices out late. Each one costs a sale or an evening.
