I find where AI helps, and make it work for the people who use it.

I don’t use AI to speed up a single task. I use it to redesign the whole workflow.

The agents I built write the first drafts of outlines, modules and job aids. But faster drafting is worthless if review becomes the bottleneck, so I redesigned the review at the same time. People decide what ships. A subject expert can reject or return any draft, and the learning team checks every course against a rubric before it goes out. Across 11 courses in 2026, expert review per course went from 90 hours to 20.

How the agents work

Every organization in the network chooses its own training, so the courses have to be worth choosing.

The hard part of organization-wide learning is rarely the platform. It’s filling the shelves with what people need, quickly. I let AI agents turn an organization’s own frameworks and standards into first drafts, so experts spend their time deciding instead of writing. At United Way Worldwide, that’s how I filled a relaunched learning platform with 30 courses in English and Spanish. Once agents were drafting, a course took 4 weeks instead of 12.

I design AI training for better decisions about when to use AI, not for more AI use.

At United Way, staff had AI tools before there was an AI policy. I designed a ten‑course curriculum for them, and the first course is built around a use guide that asks two questions of every task: what am I giving the tool, and what happens if the answer is wrong? I make the safe path the fast path. The guide is designed to clear low-risk work quickly and send only consequential decisions to formal approval, and I count over-caution as a failure too: the target is that no more than 10% of safe tasks get wrongly blocked.

The curriculum and use guide

Workshops taught me that good ideas stall without support to carry them out. So I measure finished plans, not satisfaction.

I build leadership programs around a real problem from each leader’s own organization. They leave with something built, not just ideas. At NuVuX, leaders at Rolls‑Royce, Balfour Beatty, Plexus and Peak Scientific took theirs to a working prototype. At United Way, I led onboarding for new CEOs, and each one wrote their own first-60-day plan. Of the 216 who completed it, 95% finished that plan.

See the programs

I build AI that drafts quickly and leaves the final say with the person using it.

Alf is planning software for project-based teaching. A teacher fills in a short form and gets a full unit. She can edit anything in place. It notices the edits she keeps making, and uses them in future drafts only if she switches that on. I design and build it by directing AI coding agents from written specs, with code review and an automated test suite. As of October 2026, all 1,131 tests pass.

The product storyOpen Alf