The Most Exciting Tech Moment Since My 2400 Baud Modem

Trying GPT 3.5 for the first time was the most exciting tech moment since I fired up my 2400 baud modem and logged onto a BBS.

I know how that sounds. Every tech bro with a LinkedIn account is publishing breathless takes about AI changing everything. But I’m not talking about the future of work or the disruption of industries. I’m talking about a feeling. That specific feeling of discovering something so new that nobody knows what the rules are yet.

I hadn’t felt that in twenty years.

The Misfits Era

Back when I got into online marketing, we were all misfits and weirdos. Internet enthusiasts writing about our lives, building businesses in spare bedrooms, figuring out how to rebuild marketing without the Mad Men playbook.

The tech stack was a mess. Janky PHP scripts held together with duct tape. Servers that would crash if you looked at them wrong. I watched Google Analytics emerge from server logs — first Urchin, then Hitbox, then the real thing. (And watched it crumble twenty years later with GA4, but that’s a different rant.)

We scoured Overture for one-cent clicks. We fought constant battles: SEO algorithm changes, email deliverability, shifting platform rules. Nothing was settled. Nothing was professionalized. Nothing was safe.

It was a fun time to build.

Eventually, patterns emerged. Things stabilized. SEO became a discipline. Marketing automation became a category. The wild west got fenced in. And somewhere along the way, the misfits became professionals. Some of us even put on pants.

The Conference Room Era

People like me inevitably moved to management.

One day you’re in internet marketing forums arguing about PageRank. The next you’re in conference rooms arguing about headcount. You build teams of increasingly specialized people — the SEO person, the paid acquisition person, the email person, the analytics person. Each one deep in their lane, expert at their slice.

The building stopped. The meetings started.

I spent fifteen years climbing from individual contributor to VP of Marketing at Automattic. Big teams. Real budgets. Conference rooms in multiple time zones. The work was meaningful. The people were smart. But I wasn’t building anymore. I was managing people who were building.

And I worry about what’s coming for those specialized folks. The people who became really good at one thing — at running Facebook ads or managing email campaigns or building attribution models. The ground is shifting under them, and most of them don’t feel it yet. The ones who do feel it are scrambling to figure out how AI fits into their specialty. That’s the wrong question.

The BBS Moment

Then GPT 3.5 dropped.

I can’t fully explain why it hit me like it did. The output wasn’t even that good. But something clicked. That energy came flooding back — the same energy I felt the first time I connected to Infinite Space BBS in Orlando, back when watching ASCII text scroll across a screen felt like magic.

The only thing I knew for certain: I wanted to participate.

Not from the sidelines. Not as a manager overseeing other people’s experiments. I wanted my hands on the thing.

So I left. December 2023. Here’s what I wrote on my way out:

“The web has always been a good place for builders, but the combination of the modern web stack paired with all the recent AI improvements makes this moment the best time in history for an individual person to build a nice web business. I intend to return to my web hacker roots and build something awesome.”

I meant it.

The Constraint

I set one ground rule: do everything myself.

No employees. No contractors. No specialists handling their specialty. Just me and whatever tools I could wield effectively. That constraint forces you to actually use AI well — not as a toy, but as a force multiplier.

Here’s the thing most people miss: starting early, when the models were bad, was an advantage.

When models were unreliable, you learned feedback loops. When context windows were tiny, you learned the real cost of bloat. When outputs went sideways at the slightest provocation, you learned precision. You learned to stay close to the model, to avoid abstraction layers, to develop intuition for what these things actually do well.

The people diving in now with Claude and GPT-4 don’t have that foundation. They’re working with tools so capable that bad prompting still produces decent results. They’ll plateau fast and not understand why.

I’m not just writing code with AI anymore. I’m using it for research, planning, marketing, analysis, reporting — the full stack of running a business. A year in, I have opinions about what works and what doesn’t.

Why I’m Writing

Developers have been great at sharing their practical AI work. Experiments, documentation, real stuff. Actual things they built and how they built them.

Marketers? Fluff. Endless fluff.

Thought pieces about “the future of marketing in the age of AI” that add nothing. Hot takes about jobs disappearing that offer no path forward. Content that exists to position the author as a thought leader rather than to help the reader do anything useful.

I want to bring the builder approach to marketing content. Less theory, more practice. Less “here’s what might happen” and more “here’s what I did and what I learned.”

Sometimes I’ll be wrong. That’s the point. You can’t figure things out from the conference room. You figure them out by building things, trying things, breaking things, and writing about what actually happened.

The misfits and weirdos are back. The rules are being rewritten.

If you want to watch it happen — or better yet, participate — stick around.