Tony Sharon

Independent AI systems practice · St. Paul, Minnesota

Custom AI tools, built like products

Tony Sharon is an independent AI systems integrator, a fabricator, and the founder of Pathfinder Prototyping in St. Paul, Minnesota. He builds custom AI tools, automation, and agent systems for small businesses and working professionals, and he builds them the way he builds products: planned in writing, tested against real data, and handed over with the source. He came to software from the shop floor, by way of CNC machining and hardware development, and he runs the practice alone. Since April, that one desk has put 44 products through Elgato review and onto a public marketplace. That is the standard a client's system is built to.

Tony Sharon, studio headshot in a charcoal suit and navy tie, looking straight at the camera.
Tony Sharon · St. Paul, Minnesota
45 plugins built 44 live on the Elgato Marketplace, 1 in review 5,484 commands across the line 18 releases published in one day 25,434 photos renamed by AI on his own GPU 37 certificates in 11 months as of October 2026

Three reasons to hire him

Manufacturing and AI, both first-hand

Tony came to AI from manufacturing. He ran Hurco and Haas CNC machines, wired LED signage, and cast resin parts before he wrote production code, and his practice now builds RAG document pipelines and custom MCP servers, the connectors that put an AI model inside the software a business already runs. A client gets someone who understands the work on the floor and the system that automates it, and who has been paid for both.

Pattern first, then a product line

The first Stream Deck plugin shipped in April 2026. Rather than build the second from scratch, he separated the engine from the application, and that engine now carries 45 plugins and 5,484 commands across CAD, PCB design, laser control, 3D printing, and creative tools. It is how he approaches every build: do one well, isolate what generalizes, and reuse it. For a client, it means the first system is built to become the second.

A one-person production line

In a matter of months, one desk produced 45 products, a company website, a working business plan, and a run of commissioned physical builds, and the products are in public for anyone to judge. AI did the multiplying. What he brings is the judgment that steers it: what to build, what done looks like, and what to leave out. That judgment is what a client is hiring.

What is actually for sale

Time. Every plugin in the catalogue exists because a command sat three clicks deep and somebody needed it two hundred times a day. Every dial with a live readout exists so nobody has to get up and check. The state keys, the follow modes, the dash on the strip when the app is closed: all of it is one product, a few seconds handed back, over and over, until it adds up to an afternoon.

Custom work is held to the same standard. The tool has to give back more than it cost, the quote is on the outcome, and the client walks away owning it. Good, honest business is not a slogan here. It is the only kind that produces a second engagement, and a second engagement is how a one-person practice grows.

Because in the end, what every client is really buying is

a bit · more · time