The method stayed the same. What holds the context changed.
Every service runs three ways at once: what the rules say, what the staff actually do, and what the software lets anyone do. The problems live in the gaps. I find the gaps, build the fix as working software, and make the system show its work — who decided what, and why. I did it by hand for twenty years. Now AI holds the context. A person still signs every decision.
How it goes, every time. I find everyone with a stake and interview them. I show them what it could look like as something they can click, not a deck. I work out where the new system takes over from the old ones, and where it hands the decision back to a person. I price the options and send the numbers up without my thumb on the scale. Then I measure what shipped against the process it was supposed to serve, and keep a person on the hook for it.
Each engagement moved the method one layer closer to the machine.
- 2006–13Mandarin Oriental Hotel Group. Where it started. Every executive with a stake, interviewed. The future shown in design. Costs and value sent up as data.
- 2013–19League, my consultancy. The same thing with a team: five core people, specialists for the big projects, everything through me.
- 2019–21Jitjatjo, now part of Dayforce. The pay rules written into the platform. One seam every product calls through. An audit tool that proves the rules hold.
- 2022Electriq Power, via Accella. The system recommends, the operator decides, every event reconciles. Shown to investors as working software.
- 2024–A federal childcare benefit, via Vizr AI. The intake rebuilt from the coordinators’ own checklist. AI agents that only read what they’re allowed to. Tests written before the build. A person signs.
I work through my firm, Vizr AI. I’m in Charleston, SC, and I work remotely with teams anywhere in the US.