For my first blog post, there’s probably no better topic to discuss than AI and its role in software and web development. I’ve coded hundreds of thousands of lines of code in twenty-plus years of working on websites and software, and AI is truly a sea change for anyone who does software development. It takes work that used to be tedious, detail-oriented, and time-consuming, and plows through it better than I could have in a million years.
It also may not be able to count the r’s in strawberry. It hallucinates solutions. It takes you down paths you didn’t intend.
But it has its place in the engine room for any development shop. I use AI for the mundane stuff, so I can focus on user experience, on what problems I’m solving and how best to solve them, instead of fighting misplaced colons or documenting database architecture. It’s a computer doing work a computer was designed to do. Using a stone to break open a nut didn’t make us less human. In fact, the opposite is true.
Using tools so we can focus on helping people is fundamentally human.
- Stone
- Ox
- Tractor
- AI
Nothing I ship is just whatever AI spits out
I use AI to help structure and write code, and mock up ideas and first drafts faster. But there’s still the same process of many iterations, revising line by line, reviewing what assumptions the AI is making. I’m not the kind of person to set a million agents running, not review what they do, and ship whatever they built. If I did that, I wouldn’t be doing a very good job. Doing so would be shipping AI slop, and it’s what most of us have come to despise.
The risk
AI used as a replacement for critical thinking, for crafting narrative, for creative thinking, strips out human functions we all crave to perform, and crave in the products and services we pay for. You can see it in AI slop. Left to its own devices, AI comes up with the most generic, lowest-common-denominator output. The goal shouldn’t be the lowest common denominator. The goal should be to tell the best story, to solve the problem elegantly, to enable us to spend more time on connecting with others.
There’s also huge environmental and economic risk, and like most, I’m quite concerned about it. It’s one reason I only use Claude and Anthropic’s products: I believe they’re attempting to forge a responsible path forward. It’s concerning how little has been done, as of September 2026, to regulate AI. There is still no comprehensive federal AI law in the United States, just a patchwork of state rules. Without government intervention, the only responsible thing to do is set our own restrictions.
The restrictions I hold myself to
Nothing is trained on my clients.
I’m essentially renting Claude’s brain. Anthropic doesn’t train on my or my clients’ information, and certainly nothing related to personally identifiable information.
AI doesn't speak for me.
It’s useful for drafting outlines or refining my own words, fixing grammar, making editing suggestions. The words stay mine.
AI doesn't produce visual work end to end.
Color, spacing, fonts, what goes where: those are still my decisions. AI just helps me build it way faster.
AI doesn't generate real people or real places.
We’ve all seen the commercials with the AI dog. They give me the ick, and I think that’s true for everyone.
Everything is honest.
AI makes faulty assumptions and miscalculates figures. If a customer feels misled because AI misrepresented your pricing or your product and you weren’t intentional about its output, that’s an authenticity problem. We should know what was AI-written and what wasn’t, what was AI-generated and what wasn’t.
What this means if you work with me
In my full-time job, I’m multiple times more productive than I was in the ten years prior. That’s part of why I started New Rock Digital. I’m uniquely positioned to leverage AI as a developer: I can connect it to systems, I know how to code, troubleshoot, build interfaces, and iterate. I leave the technical tedium to AI. I don’t need to manage the tiniest details of a server migration or write every line of media queries. I focus on what I’m actually trying to build and what the experience is like for users.
For a client, that means:
- One person can deliver what would take a much larger development firm, for much, much less money.
- What you're paying for is my knowledge of how to leverage AI to deliver solutions.
- You get a product that competes with enterprise products in function, performance, and security, at the cost of working with a small local shop.
Small businesses can compete with institutions that have vast resources, because intelligently leveraging AI levels the playing field like no other technology in human history.
Walking the line
AI isn’t all inherently bad, it’s transformative. But it’s up to us to determine the nature of that transformation. There’s a fine line we all need to walk in the coming years. I’m excited to try and walk it.
