journal / AI, systems design and software

AI helped me cross the distance between systems design and software.

I became Countercult’s CTO by working from product design into software until the system needed someone accountable across all of it. AI changed what I could build directly. It did not decide what Countercult should be or make me responsible for less.

01focus

The role grew out of the product

I did not start with a plan to become a CTO. I grew up inside West Point, passing through its gate guards as part of everyday life, then worked alongside engineers at technology startups as creative lead and CMO. I was building Countercult around a problem I understood from cultural work: people, places, knowledge and opportunities were difficult to find and connect. As the product took shape, the work moved from experience design into data, APIs, software and engineering decisions. I stayed close to those layers because they determine what the product can actually do.

02proof

Spatial software helped me learn the technical side

I had already learned to work with complex spatial and design software. Blender is my favorite, and I have also worked in Rhino. They are not the same as building a web application, but they made technical work feel familiar: each tool has its own logic, controls and ways of getting from an idea to a result. Once you know one instrument, it becomes easier to learn another. I brought that adaptability to systems questions in Countercult. What does a person need to do next? Which information should be public, and which belongs to members? How should discovery connect to participation without exposing relationships that need protection? The Network, Exchange, Studios and Fund have different purposes and boundaries. Designing those relationships is product work, but it is also architecture. Before I could use AI to build at speed, I had to learn the software, open-source tools and stacks I was working with, and understand how they fit together.

03proof

Vibe coding worked after I learned the stack

Once I understood the software, open-source foundations and stacks behind Countercult, I could use vibe coding to its full potential. I knew enough to direct the tools, recognize when their suggestions did not fit the product, and keep the system coherent as I built. Before AI, much of that implementation meant manual labor. AI took away a lot of that labor and let me build extremely rapidly with almost no overhead costs. The speed came from combining what I had learned with AI-assisted implementation, not from handing the architecture to a tool.

04proof

Vibe coding can start a product, but it cannot carry the responsibility

Prompting an interface into existence can make early experimentation faster. A working screen is not yet a dependable product. Someone still has to decide how the data is structured, how one part of the system connects to another, who can access what, what happens when an assumption is wrong, and how the software can be maintained. AI can help me work through code. It cannot take responsibility for the consequences of the system.

05proof

Adaptability matters more than memorizing a stack

The software changes as frameworks, APIs and product needs change. My job is not to perform expertise by remembering every command. It is to understand enough of the system to ask useful questions, read what the tools produce, find where an answer breaks the product logic, and verify the result. Design helps me keep the whole experience in view while I work through the implementation.

06proof

What a hacker residency could make possible

I am interested in what hacker residencies could become when AI makes it easier for people with different backgrounds to build software. The useful version would be more than a short-lived demo or a room full of prompts. It would give people time to learn together, work on real systems, share technical knowledge and ask who a tool serves. The strongest work would leave behind something understandable and supportable, not just something that looked finished on the last day.

07next

AI changed my reach, not my responsibility

AI helped me take on software work that would have been much harder for me to do directly. My CTO role comes from connecting product decisions, systems design, software and the needs Countercult was built to serve. I still have to understand what we are making, decide what belongs in it, and stand behind what ships. That is the difference between using a tool to build and handing the tool authorship of the system.

continue through the record

Read the Countercult product design and founder case study ↗Read how building Countercult changed my approach to data and design ↗Read my article on artists, AI and authorship ↗Read the personal background that shaped how I build ↗See selected work ↗