Philosophy
Culture drives AI transformation. Not the other way around.
Most AI rollouts fail the same way: tools arrive before trust.
Leaders mandate usage, urgency is manufactured through scarcity, and teams comply without conviction. This means nothing actually improves.
I'm not interested in tools for their own sake. Instead, I'm eager to bridge the gap between where work is and where it could actually be: scoping automations that are worth building, using Cursor and similar tools to build things myself, and finding new ways of working that actually stick inside an organization.
What I've built
GoofyScoops
Personal project · Next.js · Supabase · Vercel · Cursor
Built a mobile-first household tracking app for our dog Penny: logging food, supplements, and medical notes so our household stays in sync. The most impressive artifact on this page isn't the tech stack: it's that it's deployed, used daily, and actually sticks. I built the first version with Cursor in less than a day and iterated from there. Long list of feature requests (mostly from my wife and cousin) to be delivered shortly!
Outreach CRM
Personal project · Airtable · Zapier · Gemini · prompt engineering
Built a system that takes outreach message info, structures it with an Airtable clipper, then uses a Zapier + Gemini pipeline to generate a personalized first-draft cold email. I still edit and send every message: it matches my assets to prospect needs, generating something tailored enough to require few if any revisions. The key design constraint: AI in the loop, human hand on the wheel.
Content quality tool
Work project · ML scoping · operationalization · handoff design
Scoped and handed off a ML pipeline that scores podcast transcripts for content quality — primarily for metadata enrichment. A common statewide pattern where ML is used to operationalize a relatively manual step. The interesting part wasn't the model: it was scoping what ML is actually good at, designing the handoff with the data team, and building the person-in-the-loop controls so editors could trust and override it.
Also built with Cursor
This portfolio site — bootstrapped with no external dependencies beyond Next.js, deployed on Vercel.
Conviction coaching site — a new approach I took to reach more accessible team coaching.
What I'm still figuring out
The CRM works, but enrichment is still manual. I'm watching Clay and similar tools to understand when that part of the stack becomes worth the investment.
On an organizational level, I've spent most of my AI time at the workflow level — personal pipelines, small collaborations, small team experiments. I'm watching Day and similar tools to understand what the part of the stack becomes: personal pipelines, small collaborations, small team experiments. I'm excited about the potential of implementing AI for organizational processes. PSA governance, having a team of AI tools move from adapters to deployers. Frankly, I don't know what I'd use that at scale yet.
That's not a disclaimer: it's why this kind of work is interesting to me. Having spent over a decade at the intersection of people, process, and technology, I understand why the judgment I've developed about what makes chang stick matters more than a checklist of tools I've administered.