Start with the constraint, not the tool.
Somewhere in your business, someone is doing the same task by hand fifty times a week. Somewhere else, an AI pilot you already paid for sits on a shelf. The first step is a diagnosis: map where the time and money actually go, then find the few places where automation genuinely pays. Sometimes the fix is AI. Sometimes it is killing a handoff. The diagnosis decides, not the tool.
Working automation, in production, on your systems.
I build the thing, not a strategy deck about the thing. Claude-driven document pipelines, agent workflows, review gates, and the unglamorous glue between the tools your team already runs. No new platform to babysit, no ML team to hire.
Adoption is a learning problem, not a software problem.
This is the part most AI projects skip, and the reason most of them stall. I spent thirteen years in instructional design before anyone called this work AI. Every build ships with what your team needs to run it without me: documentation, job aids, working sessions with the people who own it. When I leave, the system is yours. That is the point.
Built, shipped, and running today.
21 years at Intuit
Staff program manager across TurboTax and enterprise learning. My last two years: building agentic AI systems used by a 20,000-person customer service organization.
Production SaaS
Lead developer of ConvoWize, an AI conversation-coaching product built with two partners and live today, from architecture through billing.
Client work in production now
Document automation for a commercial real estate firm. An AI course-development engine for an e-learning vendor. Paid engagements, deployed, owned by the client.
Small first step. Value before commitment.
Diagnostic
A short, structured look at where your team's time actually goes. You get a written map of the highest-payoff opportunities, useful whether or not we go further.
The quick win
One workflow, fixed scope, fixed price, shipped to production. Priced on the value it returns, not the hours it takes.
Expand if it earns it
Once the first win is running, we decide together what deserves the next build. A retainer when it makes sense. No lock-in.
Tell me about the task your team hates.
Two sentences about the work that eats your week is enough to start. If I am not the right fit, I will say so and point you somewhere better.