I think like an architect. I always have. How technologies connect, how systems work together, where speed and efficiency live without cutting corners. I’ve been building on the web since the beginning, and I’ve watched every major shift in how it works. When AI tools arrived, I went in immediately, the way I always do with new technology. But this was a different scale entirely. Nothing I’d seen in three decades came close. I figured out early how to do simple, high-impact things with it. Then I kept building. The tools got better, I layered on new skills and uses, and each advance opened something new. Claude Code and Claude Cowork have been the highest-impact tools for me so far, but the foundation was everything I’d already learned about how to use AI well.

Where it started

In 1995, I was hired as an art director at a catalog company. The CEO was a forward thinker. He said we needed a website and asked me to figure it out. There wasn’t a book on how to write HTML at the time. I learned by viewing source on other websites and figuring out how it worked. Coding didn’t scare me. I’d bought a Commodore VIC-20 as a kid and taught myself BASIC. HTML was just the next thing to figure out. And editorial wasn’t foreign either. My first graphic design job was at a newspaper, where I laid out pages by hand. I worked alongside editors and journalists every day, learning how they selected stories for A-1, chose photos, crafted headlines, and shaped the edit. That process stuck with me.

The CEO’s goal was to drive catalog requests online. That was a great idea, but it needed a hook: a reason for people to come to the site in the first place. I read an article about a brand that wrote content on topics they knew their customers loved, unrelated to the product, to attract people to the business. Content marketing, years before anyone called it that. I hired a journalist and launched an online magazine. Editorial integrity was the rule from the start: strong stories about the topics our audience cared about, never advertorials. The catalog request lived outside the article body. My rule: Write something worth reading, and the business gets exposure on its own terms. It worked.

Then I found Viaweb, the e-commerce platform Paul Graham built. I went to the CEO and said, “This is the future.” I launched a store featuring 50 of our most popular products from our 10,000-SKU warehouse and promoted them through editorial content. Both sides were working. Catalog requests from the web were generating a higher average order than the print catalog, and the store continued to grow. I ended up running the number one store on Viaweb for traffic and revenue before Yahoo! acquired the platform.

Around this time, I started sitting in the circulation director’s office. He showed me how catalogs were tagged to assess list value, how you adjust when you’re mailing too many catalogs to people who don’t buy, and how every decision was driven by data. I knew web stats existed, but I hadn’t looked at them the right way. Watching how data-driven the catalog business was opened my eyes: I could be doing the same thing on the web. Which articles were performing, which topics the audience was voting for with their clicks, and how to shape editorial around what the numbers said. That connection wasn’t obvious to most people at the time. It was to me.

I wanted the editorial and commerce sides to talk to each other. When I scoped the build for a custom platform, I designed a shared taxonomy backend. An article about The X-Files would automatically surface the most relevant, most popular, and most profitable X-Files products. When we launched it, conversions jumped. I called it contextual commerce. That was 1997.

By the time I left, the company had grown its annual revenue from $12 million to $20 million, with the web projected to generate $8 million that year. I’d gone from art director to building the systems that drove most of the growth. And I’d found the work I’d be doing for the rest of my career: figure out how content, taxonomy, and systems connect, and build the thing that makes them work together.

Squidoo

I co-founded Squidoo with Seth Godin in 2005 and served as Chief Product Officer. I wrote the 100-page scope document that defined the platform, selected and hired the agency to build it, and oversaw development through launch. We grew it from zero to 82 million monthly visits, entirely organic. It was one of the first user-generated content platforms on the web, and it paid more than $18 million to members and charities. It was acquired in 2014.

I’ve led platform strategy and web development for clients across media, retail, education, and nonprofit. Chico’s FAS recruited me after Squidoo’s acquisition and brought me in to work with VPs across their business units on web strategy, systems, and problem-solving. A few days in the office turned into a year-long consulting engagement. The work is always the same at its core: figure out the content architecture, build the systems, and make it scale.

No Treble

I founded No Treble in 2009 because the publication I wanted to read didn’t exist. It became the leading online magazine for bass players, reached 500,000 monthly visitors at its peak, and ran for 16 years before it was acquired in 2025. I designed the editorial model from day one to build community, not just audience: readers featured alongside legends, curated videos that turned unknown musicians into overnight discoveries, and a sticker campaign that became a badge of belonging. When the pandemic threatened the site’s survival, I put the decision in the community’s hands. They raised $22,000 in three weeks.

I’ve been playing bass since I was 10. That’s not a side note. No Treble exists because bass is central to my life. The platform’s credibility came from deep knowledge of the subject, not from market research.

Then AI arrived

I signed up for ChatGPT in late 2022 and started using AI seriously in my work by early 2023. It clicked fast. Not because the tools were magic, but because I had decades of context to bring to them. I knew what to ask for. I knew what good output looked like. I knew when it was wrong. Everything I’d spent three decades building turned out to be the foundation for using AI well.

At Muse Group, where I lead editorial for Ultimate Guitar and MuseScore, I started applying AI to real problems. I built an editorial operations platform with Claude Code: a centralized dashboard for workflow management, prompt generators, trending topics, one-button WordPress publishing, and team communication. It replaced scattered tools that weren’t built for how the team actually works.

I also built an NLP-based headline-grading tool that delivered a sustained 75% increase in traffic during an A/B test. And I developed AI-assisted content evaluation prompts that automate E-E-A-T assessment, SEO review, and fact-checking with ranked feedback. The prompts were developed through extensive data analysis and testing, and managed centrally so authors never touch them. The tool pushes content with full HTML so the LLM can evaluate citations, links, and structure.

But the part I’m most proud of is what happened with my team. They were skeptical of AI. Some were resistant. I didn’t mandate adoption. I worked with each person individually, built tools that protected their voices and expertise, and showed them how AI could improve their work without replacing what made it theirs. They went from skeptics to advocates. When leadership suggested I run a company-wide AI workshop, colleagues immediately echoed that they wanted the same thing. I ran a 90-minute session and opened with this: “You’re here for a reason. Your experience, your knowledge, your expertise. AI doesn’t replace that. AI in your hands is more powerful than AI in someone else’s without your experience, and more powerful than AI on its own.”

Putting people at the center isn’t a talking point for me. It’s how I build tools, run workshops, and lead my team. I put expertise first and build AI around it.

Outside of work, I built Bassists.com with Claude Code, scoping the entire project the way I would for a development team, then building it in a weekend. Every discography is curated by me because I know which albums a bassist actually recorded on, and no API is reliable enough to get that right. AI handles the mechanical work. I handle the judgment. I wrote about the process in How I Built a WordPress Site with Claude Code in a Weekend.

The thread

Content platforms, editorial systems, taxonomy, community, and AI. These aren’t separate tracks. They’re the same work, refined over time. Every project taught me something that made the next one better. AI is the most powerful tool I’ve encountered in three decades of building, but it’s only as good as the experience behind it.

If you want to see how I think about AI in practice, start with my Working with AI posts. If you’d like to connect, reach out.