Talk and diagnose
I sit with the team and watch the work, in person in Louisville or by video anywhere else. I map where the time goes. I just want to make sure it's something you'll actually use before we commit to building it.
Services
There's a long way between trying ChatGPT and having AI run a real part of the business. This page is how I get a business across it, one step you approve at a time.
The Census Bureau's business survey puts AI use at roughly one firm in five. Private surveys of small businesses put it anywhere from 66% to 76%. A lot of the distance between those numbers is the distance between a login and a working system.
A vague problem is how an automation project gets expensive. If nobody has written down what's being fixed and what it should save, the scope grows to fill the budget. The other tell is a tool named in the first meeting, before anyone has watched the work.
If anything changes, we agree it in writing before it changes.
It runs on your own process before there's any talk of a full build.
I sit with the team and watch the work, in person in Louisville or by video anywhere else. I map where the time goes. I just want to make sure it's something you'll actually use before we commit to building it.
A small, fixed-price working prototype of the one piece worth proving, running on your own process, so you watch it work before deciding anything bigger.
A Louisville recruiting firm did exactly this. The prototype pulled open job listings, researched each company, scored it against the kind of client they wanted, and drafted the outreach for their team to send. They paid for it, and I built and delivered it. Read the case study →
The production build, at a fixed price you've approved. What that usually means:
Either I keep it running month to month, watching it, tuning it and adding the next piece, or I train the person on your team who'll own it.
Some teams need a senior person inside for a quarter or two, then gone, with the system left running. At Global Payments, a 27,000-person company, I built the intake and governance engine that routed each piece of work to the right approvers by type, scope, systems and geography. It ran in production for years and outlived my 2022 departure. It was built before AI existed, so every routing call in it was written by hand.
For owner-run service businesses that want it handled as one monthly service with nothing new to learn, that's Hire Lexi.
Before a bid goes out, somebody has to read every piping drawing in the packet and count what's on it. I'm building the engine that does that read for them, as a paid build. A scanned drawing goes in, it finds the tables and labels, reads them, and hands back a material list their team can check. When it can't read a sheet, it says so instead of guessing.
Read the case study →They pay me per site. Eight sites went out in the first two weeks, built one at a time. Now it's one factory, where a new dealer is a data folder: their logo, their cities and their form. Their clients never hear my name.
Read the case study →I wrote the offer once. The engine finds who to send it to, scores whether their audience is actually a fit, and runs the whole email sequence on its own. It runs itself daily.
Read the case study →If there's a process in your business you'd like someone to look at, tell me about it.