Writing about AI adoption taught me the advice I gave in January looks different from where I landed in December.
I’ve now published 100 posts about AI.
Looking back at the past two years, the advice I gave in early 2024 looks different from where I landed in December 2025. Some calls aged well. Others didn’t. And a few things blindsided me entirely.
Here’s the honest accounting.
What I Got Wrong
Let’s start here — it’s where the real lessons are.
I was too bullish on AI agents.
In November 2024, I published “AI Agents: The Next Evolution in Marketing Automation (2025 Preview)” — painting an optimistic picture of autonomous systems handling complex marketing tasks.
Seven months later, I wrote “AI Agents Won’t Save Your Business. Better Operations Will.”
What changed? I watched companies chase the agent hype while ignoring the operational foundations that make agents useful. The technology is real. But agents don’t fix broken processes — they automate them faster. A company that can’t clearly define its workflows won’t be saved by an AI that executes those workflows autonomously.
I overcorrected toward the technology and undercounted the organizational basics.
My early content was tool-focused when it should have been systems-focused.
Read my posts from early 2024: “which tools to use,” “how to get started,” “top AI marketing tools.” Helpful at the time, but missing the bigger picture.
By late 2025, I’d shifted to “Stop Chasing AI Tools. Start Building an AI Engine.” The realization: tools are commodities. Everyone has access to the same ChatGPT, the same Claude, the same automation platforms. The differentiator isn’t which tools you use — it’s the system you build around them.
I spent too long answering “which tool?” when the better question was always “what are you trying to build?”
What I Got Right
Not everything missed the mark.
SMBs have architectural freedom enterprises don’t.
I started making this argument in early 2024 and doubled down throughout 2025. Small businesses can redesign workflows from scratch for AI capabilities. Enterprises stay trapped by legacy systems, sunk costs, and the politics of changing anything.
This held up. The SMBs I worked with who succeeded weren’t the ones with the biggest AI budgets — they were the ones willing to rethink how work gets done.
Tool sprawl would become a trap.
By mid-2025, I was calling it the Marketing Ops Death Spiral — companies adding AI tools without subtracting complexity. The pattern played out everywhere. The companies that avoided it weren’t the ones who found the “right” tools. They were the ones who stopped asking “what tool should we add?” and started asking “what problem are we actually solving?”
The skills gap would stay undefined.
In October 2024, I wrote about the AI marketing skills gap and urged people to start upskilling. Fourteen months later, “AI fluency” is the fastest-growing job requirement — and employers still can’t define what it means. The gap didn’t close. It just got a fancier label.
What Surprised Me
The hard part isn’t AI — it’s everything before AI.
This one hit me repeatedly throughout 2025.
Early in the year, I focused on prompts, workflows, automation sequences. By fall, I was writing about why marketing data should work like banking records, why perfect data is a lie, and what event-sourcing actually reveals about attribution.
The pattern: companies that struggled with AI didn’t have an AI problem. They had a clarity problem. Messy data. Undefined processes. Workflows nobody had documented. They couldn’t tell AI what to do because they couldn’t articulate what they were trying to accomplish.
The surprise wasn’t that data matters — everyone says that. The surprise was how consistently this upstream work determined downstream AI success. Clean data and clear workflows aren’t preparation for AI. They’re the whole game.
I became a builder.
This wasn’t on my 2025 roadmap. At all.
I wrote about building a webhook without coding because I had a problem and no developer to solve it. Then I built a Chief of Staff AI to stop myself from giving away free work. Then a GEO analyzer because I kept seeing clients make the same content optimization mistakes. Then a voice app to practice my terrible Dutch because… well, my Dutch is terrible.
I’m not a developer. I don’t have a CS degree. But AI collapsed the gap between “I have an idea” and “I have a working tool.” That shift — from consumer of AI tools to builder of AI systems — changed how I think about what’s possible. It also changed my advice: I stopped telling clients which tools to buy and started asking what they wanted to build.
The “AI fluency” hiring market is completely broken.
McKinsey says AI fluency demand is up 7x in two years. PwC shows a 56% wage premium for AI skills. And after working with salespeople and marketers who considered themselves AI-savvy, I found most are missing the fundamentals.
They use tools daily. They don’t have systems.
Employers are hiring for tool usage when they need systems thinking. Candidates are listing “ChatGPT” on resumes like it’s a qualification. Nobody agrees on what fluency actually means — which creates opportunity for SMBs willing to define it for themselves.
The Theme Underneath
2024 was about adoption. Are you using AI? Which tools? How do you get started?
2025 was about architecture. What are you building? What systems support it? What needs to be true before AI actually helps?
The question shifted from “are you using AI?” to “do you have a system for using AI?”
If I had to bet on what separates companies that thrive with AI in 2026 from those still spinning their wheels, it’s this: the winners won’t be the ones with the best tools. They’ll be the ones who got clear on what they’re building — and did the unglamorous upstream work to make AI useful.
That’s where I’m starting 2026. I’d encourage you to do the same.
Not Sure Where to Start?
If you’re looking at your own AI journey and wondering what’s working, what’s not, and where to focus next year — that’s exactly the kind of clarity I help SMBs and marketing teams build. Let’s talk.
Leave a Reply