In August 2025, I wrote a piece called “The Machines Are Here.” Not a blog post about chatbots. A prediction — the kind that made people in my circle think I was being dramatic.
The core claims:
- AI won’t just assist. It will replace entire workflows.
- Small towns aren’t ready.
- White-collar work restructures first.
- “Business as usual” is about to get debugged by the machines.
Eight months later, let me grade my own homework.
Prediction: “AI isn’t coming for your job. It’s already here.”
Grade: A. Q1 2026 saw 60,106 tech jobs cut across 204 companies — 668 people per day. Amazon cut 16,000 corporate roles. Meta cut up to 15,800 while simultaneously spending $135 billion on AI infrastructure. Block laid off over 4,000 people — half its staff — explicitly because AI does the work now.
The jobs aren’t “threatened.” They’re gone. The companies doing the cutting are posting record profits. The money didn’t disappear. It just stopped going to human paychecks.
Prediction: “What used to be easy is already obsolete.”
Grade: A. The $20/month tier of Claude is now more powerful than the $200/month tier was twelve months ago. Tasks that required specialized knowledge — data analysis, report generation, code development, content creation — can now be done by anyone with a subscription and 30 minutes of setup time. The floor for “competent” just got raised, and everyone below it is visible.
Prediction: “What once felt impossible will be automated next.”
Grade: A+. I didn’t predict this one strongly enough. 70 to 90 percent of the code behind Anthropic’s future models is now written by Claude itself. Not supervised line by line. The machine is writing the next version of the machine. Anthropic’s alignment lead: “Recursive self-improvement is not a future phenomenon. It is a present phenomenon.”
In March, AI went from “tool that helps developers write code” to “system that develops itself.” That’s not a productivity improvement. That’s a category change.
Prediction: “Small towns aren’t ready.”
Grade: B+. The prediction was right. The impact is slower than I expected — not because the technology isn’t there, but because adoption in small-town America is lagging by 12-18 months behind metro areas. That lag is both the problem and the opportunity. The businesses in Waseca, Montgomery, and Le Sueur that adopt now will have the same tools as companies in Minneapolis. The ones that wait will be competing against neighbors who didn’t.
Prediction: “The Church isn’t ready.”
Grade: A. Still true. Most churches are either ignoring AI entirely or panicking about it as an eschatological threat. Almost none are doing what needs to happen: helping their congregations understand and adapt to the most significant technological shift since the internet. The pastors who get this right will be the ones their communities remember. The ones who pretend it isn’t happening will lose relevance faster than they already are.
What I Got Wrong
I underestimated speed. When I wrote “The Machines Are Here,” I thought we had 2-3 years before the impact was widely visible. The Q1 2026 layoff data says it’s visible now. The acceleration curve didn’t follow my timeline — it outran it.
I overestimated resistance. I thought more institutions would push back against AI adoption. Instead, the US Senate authorized ChatGPT for staff, 81% of physicians now use AI, and enterprise vendors are building AI into everything. The resistance phase was shorter than expected.
I didn’t predict the integration layer. MCP — the protocol that lets AI connect directly to business tools — wasn’t on my radar in August 2025. It turns out the thing that makes AI go from “interesting toy” to “operational infrastructure” isn’t the model getting smarter. It’s the model getting connected. That connection happened faster than anyone expected.
What I’m Predicting Next
For the record, so I can grade myself again in Q4:
-
By Q4 2026: Companies that didn’t adopt AI in the first half of the year will be operationally behind — not strategically, operationally. Their competitors will ship with 3-person teams what used to take 15.
-
By end of 2026: At least one major professional certification body (accounting, legal, or medical) will require demonstrated AI competency for licensure or continuing education.
-
Within 12 months: The “AI consultant” role as currently understood will be commoditized. The value won’t be in knowing how to set up AI — everyone will know that. The value will be in knowing which AI to deploy for which business problem, and maintaining it as the tools change monthly.
-
The acceleration won’t slow down. It will get faster. And it will get easier. The ceiling goes up while the floor goes down.
What to Do With This Information
Same advice I’ve been giving since January: start.
Claude Pro is $20/month. Ollama is free. The knowledge to set it up is published on this blog. The predictions I made in 2025 came true faster than I expected.
The next set of predictions will too.
If you want help getting started, that’s what FIT does. We configure AI workflows for small businesses and nonprofits. We eat our own cooking — every tool we recommend, we use ourselves.
The cheat codes are still free. The scorecard speaks for itself.
Matt Stoltz is the founder of Flower Insider Technologies, an AI-assisted managed IT company in southern Minnesota. He’s been making tech predictions since before they were fashionable — and keeping score.
Full blog series and data sources at flowerinsidertechnologies.com/blog.