Michael Hoff spent almost ten years in tech, most of it on the ServiceNow platform, and ended up working as an AI engineer at a startup built around ServiceNow. Then he was laid off. In this short video he walks through the mistakes that put him there, all of which, he says, came from resisting AI tools instead of learning them, and the unglamorous skill that got him back to work within a month.
This article is our own summary, with short quotes, from the video's automatic transcript. Watch the full conversation above for everything that was said.
The short version
- He worked at an AI company and didn't use AI. Early tools produced broken ServiceNow update sets (precise XML), so he decided AI wasn't good enough for real work and stopped checking as it improved.
- He argued with the warning instead of preparing for it. When teammates said AI would get better, he answered that then none of them would have jobs. He was right about the outcome and still didn't act on it.
- He ignored the speedups. He disliked AI-written code, with its band-aids and repetition, while colleagues shipped far more features. "The business doesn't really care how much you like your own code."
- He held on to the craft. Even as a forward deployed engineer writing throwaway client scripts, the perfect case for AI, he was slow to use it. The real reason was that he loved writing code himself.
- He let it get to his head. Feeling behind made him avoid new tools, which put him further behind. His advice: treat the feeling as information, not a verdict, and talk to someone if it is affecting your life.
- What saved him was specialization. Deep ServiceNow experience is scarce, so he has never been out of work for more than a month. Pick a specialized, slightly boring platform or industry, then stack AI skills on top.
The mistakes, one by one
1. Judging AI once and never checking again
Hoff's main project turned a simple diagram of an application into a working ServiceNow application. His job was the deterministic code behind it, which he describes as a sort of compiler. When the team tried having AI generate ServiceNow update sets, the results didn't load in a live environment, so he skipped AI. He says he wasn't wrong about the tools back then; the mistake was never going back. His fix: set aside a small part of every week to try new tools, even on work where you expect them to fail.
2. Treating "AI will get better" as an argument to win
Some colleagues designed around the expectation that AI would improve. Hoff dug in. He says he treated the prediction as a debate when it was a warning, and that "not liking the direction doesn't change the direction." He should have become the person on the team who knew how to use the tools.
3. Ignoring the speedups
He was proud of code he understood line by line, but it took him a long time to write while others shipped more. Code quality matters, he says, but when a company decides who to keep, the person shipping is in a much better position than the one with the prettiest pull requests. His suggestion is to meet in the middle: get a working version fast with AI, then apply your standards in review and cleanup.
4. Holding on to the craft
He says this was the real reason behind the first three mistakes. He liked solving hard problems himself, and handing that to AI felt like giving away the part of the job he cared about. The lesson: separate the craft from the job. Write code by hand on your own projects if you like; at work, the job is solving the problem.
5. Letting it get to his head
Watching his job change, he felt his skills no longer mattered. That kept him in his comfort zone, which made him fall further behind, which made him feel worse. He calls it a loop and says feeling behind simply means you are behind, which is fixable.
What actually saved him
Not his AI skills, he says, but his "boring background." ServiceNow is specialized, employers expect real platform experience, and few people have it. By contrast, in general software work "anybody with an AI subscription can functionally do some parts of the job," which makes for a crowded line. His closing advice: use the tools, and have something specialized to fall back on.
What it means for your business
- Re-test the tools on a schedule. A tool that failed your hardest task a year ago may handle it now. Give your team a little time each week to try.
- Reward shipped outcomes. Make clear that using AI to deliver faster, with human review, is how work is measured.
- Separate throwaway work from core systems. Scripts and one-off client tasks are where AI pays off first.
- Value specialists. People who know a platform or an industry deeply, and who add AI on top, are the hardest to replace and the most useful to keep.
- Watch for discouraged people. Engineers who feel left behind often avoid new tools. Pair them with someone who uses them well.
Video details
- Title: Laid off as an AI engineer (don't make my mistake)
- Creator: Michael Hoff
- Published by: Michael Hoff on YouTube, November 2, 2026
- Watch: youtube.com/watch?v=vAzcFVSGGJM
- Note: quotes come from the video's automatic transcript and are lightly cleaned. Check the video for exact wording.
Want to put this to work?
We help owners turn AI trends into a plan, a product and a launch. Book a free call or see our services.