Background

The workshop

I grew up in Virginia Beach, mostly in my dad's workshop.

We worked on cars, did woodworking and metalworking, and always had a few projects going.

The lesson that stuck: if something isn't built the way you want it, you can build it yourself.

The workshop taught me the other side too. Projects pile up when nothing forces you to finish them. I'm deliberate now about what I start and what I see through.

I was a nerdy kid. Robotics club, jazz band, the advanced math and science classes. I built the robot, but I didn't write a line of code until college.

Georgia Tech

I went to Georgia Tech for math because I was good at it in high school. Then I got a 2 out of 100 on an abstract algebra test. I switched to computer science, mostly because I was getting 100s in the CS classes and none of the alternatives held my attention.

For a while, CS felt like a job more than a craft. I stayed for a master's, funded by a university research lab where I worked on generative algorithms for defense systems. It meant long days in a closed room with no phone and no internet. I learned a lot, mostly that I never wanted to work that way again.

The grad classes changed everything. Graphics, computational aesthetics, game design, and shape grammars showed me software could be a creative medium. For the first time I spent hours past what an assignment needed because I wanted the thing to be good.

Getting obsessed

I graduated right as ChatGPT took over the news. I'd fine-tuned GPT-2 in a class but never gone deep. With a few months left on my Atlanta lease, I started building with it every day.

When the lease ran out I moved back home and put everything else away. No video games, no golf. I learned how to wire up APIs, deploy real applications, and test ideas like synthetic polling and a few small SaaS products that went nowhere. I joined every Discord community and watched every video I could find.

Learning in agencies

Building alone only teaches you so much, so I went looking for people better than me. I chose small agencies where I'd get real build time, even if it paid less.

At Width AI I built an agent-building platform on top of FlowWise and moved from Python into TypeScript and frontend work. At Morningside AI I learned from their CTO how to actually scope and build a product, and helped build the first version of their SaaS product, Agentive.

At VRSEN AI I worked across a wide range of multi-agent client projects. I learned the full client lifecycle: sales calls, scoping, architecture, leading the build, and delivery. That was the last piece I needed to do this work on my own.

Going independent

In 2024 I moved to Austin and started Lascari AI. I'd realized I didn't want to work for someone else. I wanted to own the work, not just write code inside someone else's system.

The first year was hard. Contracts dried up, and I had to rebuild the pipeline from scratch, mostly through people I'd met during the agency years. Since then I've built a full voice agent stack, taken on advisory clients, and started writing about what I'm learning.

What I'm working on now

The thread through all of it is understanding. The systems that worked were the ones whose builders knew exactly what they knew and what the system was doing. The ones that failed were black boxes, even to the people who built them.

So now I'm building tools for that: documents, diagrams, and prompts that you and an agent both read and edit, plus an agent runtime that traces everything it does. The idea lives in one shared place instead of only in your head. Every time you fix the shared version, the AI gets better and your own picture of the problem gets clearer.