Michigan Business Owners Who Were Afraid of AI — And What 90 Days Changed

His name isn't important. What's important is that he'd been dispatching with the same method since 2011 — phone calls, a whiteboard, and a driver he trusted completely. Hiring a second dispatcher felt expensive. Software felt threatening to his guy.

We didn't replace Marcus. We gave Marcus his Friday afternoons back.

The system handles the 5 AM route build, the load matching, and the automated driver assignments — all the work Marcus was doing before the sun came up, six days a week. Marcus now focuses on the calls that actually need human judgment: the difficult clients, the unusual loads, the relationship management that only 11 years of experience can handle.

We deployed in 4 weeks. Marcus learned the system in two days. Three months later, the owner expanded from 3 trucks to 22 because he finally had the operational capacity to take on more freight without his team breaking under the load.

She ran a six-provider medical group in Metro Detroit. Her office manager had been handling prior authorizations the same way for eight years: phone calls to insurance companies, hold music, fax confirmations, manual tracking in a shared spreadsheet. Two staff members spent 60% of their week doing nothing but prior auth.

She was right. Most AI platforms are not HIPAA-safe. We don't use most AI platforms.

We build on-premise — meaning the AI runs inside your infrastructure, your data never touches an external server, and you sign a Business Associate Agreement before we write a single line of code. The system we built for her practice processes prior auth requests, tracks denials, generates appeal letters, and flags time-sensitive cases — all within a closed environment that her IT manager can audit at any time.

Her two staff members now handle 3× the case volume in the same hours. Zero HIPAA incidents. Zero. The practice applied for Going PRO reimbursement on the staff training component and recovered $4,400.

He owned a 47-person precision parts shop supplying two OEMs out of West Michigan. He had been burned before — badly. An ERP implementation that consumed 18 months, two consultants who didn't understand manufacturing, and a system his floor supervisors refused to use because it slowed them down. He was paying $3,200 a month in software licenses for a platform sitting mostly idle.

That's exactly what we did. We came in, walked the floor for three hours, and identified the one process that was costing him the most: end-of-line quality inspection. Four inspectors, six hours per shift, catching defects that sometimes made it through anyway. We scoped a pilot — AI vision system on one line, one month, fixed price.

The pilot hit 94.7% defect detection on the first run. He expanded to all three lines over the following quarter. His OEM audit in May came back with a commendation on quality process improvement. The Industry 4.0 Tech Grant covered 50% of the implementation cost.

She ran a 28-person GC operation out of Wayne County. She was proud of her estimators — two of them with 30 years of experience between them, built into the rhythm of her business. Her concern wasn't about the technology. It was about disrupting people who had made her company what it was.

The system we built didn't replace Tommy and Ray's expertise. It ran on it.

We spent two weeks documenting how Tommy and Ray thought — their pricing logic, their material assumptions, the things they "just knew" from experience. Then we trained an AI on that judgment, so every new estimate started with the institutional knowledge of her two best people, not a blank template.

Tommy called it "my brain in a box." Ray uses it every morning. The estimates that used to take three days now take four hours. She won four jobs in the first month she couldn't have bid fast enough to win before.