We’ve Been Here Before

By Karim Marucchi, CEO of Crowd Favorite and Pat Ramsey, Director of Technology. Between them, they have over 60 years of experience building technology systems across the web, enterprise software, and now AI.

Early in my career, I was working at an architectural design firm. CAD software, blueprints, the kind of work that had nothing to do with the internet. Then one day the phone rang. The voice on the other end said they needed a website built. I said, “sure, we could do that”.

It was Nissan.

We had never built a website. But we figured it out, because that’s what you did in the mid-90s. Nobody had a playbook. You brought what you knew, you learned what you didn’t, and you shipped.

Pat Ramsey, our Director of Technology, has his own version of that story.

I started my career as a print photographer. I was a photo journalist with film, chemicals, darkrooms, news layouts built with actual glue and paper. I started out completely analog and by the time I graduated from university, everything was becoming digital. Whole floors of buildings were gone as publications stopped using paste-up and mechanical processes. Entire companies built around the physical mechanics of the medium just ceased to be. Coming out of this, though, was born the Web as we know it. Graphical browsers, digital photography, application interfaces, everything around web and digital development.

Pat didn’t just watch that transition. He helped build it. On February 2, 1995, his team launched the first newspaper in Texas to have all editorial content online. Between us, we’ve watched technology reinvent itself more than once. And what strikes us both right now as we watch the AI boom unfold, is how familiar it feels. Not the technology itself. The pattern around it.

The Hype Pattern Is Familiar

AI is generating the same kind of energy the internet did. Complicated chains of AI agents sold as the only path forward. Overwhelming documentation handed over as if volume equals clarity. Change management frameworks that require their own change management frameworks.

We’ve seen this before. Most of that complexity didn’t hold up then. It won’t this time either. What survived then — and what’s winning now — is disciplined simplicity.

An Engineer in 1989 Wrote Your AI Playbook

Rob Pike, one of the architects of Unix, published five rules for writing good systems software decades ago. The short version: don’t guess where things break, measure before you optimize, and most importantly — data dominates.

Pike’s argument was that if you structure your data well, the logic becomes self-evident. The algorithm almost writes itself.

Factory.ai, which evaluates codebases for AI-agent readiness, recently published findings that are a direct echo of this. Their consistent finding in production: the agent isn’t the broken thing — the environment is. Clean linter configs. Documented build systems. Structured dev containers. When those fundamentals are in place, agent behavior improves dramatically. When they’re not, no amount of model sophistication fixes it.

What the 90s Actually Taught Us

The early internet didn’t reward the most complex architectures. It rewarded the teams that understood the primitives deeply — TCP/IP, stateless HTTP, relational databases — and built on them cleanly.

But there was another force that shaped that era just as much: open source. Linux, Apache, MySQL, PHP, the entire LAMP stack that powered the early web was built on shared, transparent, community-owned foundations. WordPress, TYPO3, and the CMS ecosystem that followed were built the same way: open, extensible, and improved by thousands of developers working in public. Nobody had to reinvent the server. Nobody had to reinvent the database. You built on what the community had already proven, and you focused your energy on what was actually unique to your problem.

AI is following the same pattern. Enormous amounts of the foundational models, the frameworks, and the tooling are open source, auditable, and improving in public. The teams winning right now aren’t the ones trying to build everything proprietary from scratch. They’re the ones who know which foundations to trust, how to build cleanly on top of them, and when to contribute back.

The companies that got buried in the 90s weren’t out-competed by smarter engineers. They were buried by their own complexity of spaghetti data structures, undocumented systems, and a refusal to build on shared foundations.

One thing we keep coming back to: if we want agents to do better work, we have to be less lazy about the environments we give them. Don’t guess, measure first, simple beats fancy, simple avoids bugs, data dominates.

All five rules. Still winning in 2026.

What This Means If You’re Building Right Now

At Crowd Favorite, this is exactly the work we’re doing with clients, and it starts in a place most people don’t expect: your environment, not your AI tools.

Before you evaluate your AI strategy, evaluate your foundations. We regularly find that bad data structures and messy infrastructure are the reason AI investments underperform. Cleaning that up is fixable, and fixing it compounds fast.

From there, we help teams identify where AI can create immediate, measurable impact, whether that’s automating workflows, accelerating development cycles, or building client-facing tools, and we build those solutions on the simplest architecture that can actually do the job — because well-architected and simple is the hardest thing to build, and the most durable.

The teams we’ve seen pull ahead aren’t the ones who spent the most on AI. They’re the ones who did the “boring” work first: clean environments, clear documentation, honest measurement. That’s the foundation everything else runs on.

We’ve Been Here Before – And That Excites Us

We’ll be honest: this moment feels electric.

Years ago, when Disney ABC Press needed to consolidate six television network sites into a single unified platform, the conventional wisdom said WordPress wasn’t up to it. It was too simple, too “bloggy,” not built for enterprise scale. We disagreed. We built a custom WordPress solution on clean, open-source foundations that manages over 14 million media assets, survived three major business reorganizations, and delivered a decade of return on investment without a full rebuild.

The technology was debated. The principles weren’t.

• Build on proven open foundations.
• Keep the architecture clean.
• Let the data structure do the work.

That’s exactly what we’re bringing to AI now. Not because we’re chasing the wave, but because we’ve ridden one before, and we know what it takes to still be standing when it settles.