AI Was Supposed to Obsolete Me. It Multiplied My Opportunities.
At 66, hiring filters rejected me in hours and never saw a person. I am not bitter about it. The same tools that score us out multiplied the range of problems I can solve, and that is worth more than any job I applied for.
TL;DR
Over eighteen months I applied for about ten jobs. Each time an applicant tracking system, or a recruiter behaving like one, rejected me within hours. I am 66, and the filters saw dates, not a person. I am not bitter about it. The same class of tools that screened me out has expanded the field of problems I can solve by ten times or more, and that has replaced the “my moat just evaporated” anxiety with something closer to confidence.
Over the last eighteen months I applied for about ten jobs. I did not apply randomly; each one was a role I could do well. The response was the same every time. A few hours after I submitted, an automated note came back: “You are not qualified. Thanks for applying.” No conversation. No chance to show the work. LinkedIn recruiters were no better; they run the same filters with a human name attached.
I am 66. The systems saw dates. They never saw a person.
I am not bitter about it. This post is about what I found on the other side, because I know there are people reading who are standing where I stood.
What the filters were scoring
An applicant tracking system is a measuring stick. It scores what it can see: job titles, years, gaps, the recency of a skill name. It cannot see judgment, because judgment does not fit in a field. So it does what every wrong measuring stick does. It rewards what it can count and discards what it cannot, and the people with the most of the uncountable thing are the first ones out.
I have written about this pattern before, in a different setting: when the measuring stick lies, good people stop trying. I did not expect to be the case study.
What I build now
I run what I would call a software factory, and I want to be precise about that word, because it is easy to hear “AI coding” and picture someone typing wishes into a chat box. I spent much of my career writing embedded software for medical devices, where a defect can reach a patient. The work I do now is every bit as rigorous. Every product has a specification. Every rule in the specification has a row in a decision register saying when it was decided and why. Every check that guards a deliverable was broken on purpose once, to prove it fails when it should. A deploy that fails any gate does not ship.
Inside that factory, this year, I replaced a 28-year QuickBooks subscription with an accounting system I own. I built a multi-project Critical Chain scheduler. I ship a paid audit product that checks whether a website is ready for AI agents, and last week I added a scoring engine to it. Three days ago Google Gemini looked at my company’s website and called it a textbook example of how a business should prepare for agentic commerce.
AI coding agents do the reading and the typing. I work inside Claude Code, Codex, and Grok Build all day, and I have written about what that is like. They read documentation I would never finish. They write the tests and do not get bored. I supply the intent, the specification, and the judgment about whether the result is right. That division of labor is the whole story.
The part most people are missing
Here is what changed my outlook, and it has nothing to do with typing speed.
Over a career I learned as much about problems as about code: which problems matter, and, more valuable, which problems people desperately want solved in some narrow or niche domain without any idea that a solution is even possible. A clinic’s scheduling. A print shop’s capacity. A contractor’s quoting. Nobody builds software for those, because the market for each one is too small to justify a team.
That constraint just fell. When one person with judgment can specify, build, test, and ship a rigorous tool in days, the field of problems worth solving expands by ten times, maybe a hundred. All the problems I have watched people live with for decades are suddenly in reach.
That is inspiring, and it has reduced my anxiety more than anything else in years. When the models first arrived, the feeling was “my moat just evaporated.” What replaced it is a different kind of confidence. The moat is knowing which problems are worth a solution, and that just got more valuable.
The commercial side confirms it. In one two-week window this summer, AI assistants fetched my pages about 1,490 times on behalf of people asking questions I answer for a living. No job board ever sent me that.
If this is you
If you are in your 50s or 60s and feel stranded at the intersection of age and AI: the market and the models did not take your judgment. That is the part the filters cannot score, and it is the part that compounds with these tools. I have been writing code in one form or another since 1977. That much of it is the scarce input in this market.
So start small and start now. Pick one problem you already understand, one you have watched someone live with. Pick one tool that helps you build the solution, and build a version that works. Then show it to one person who has the problem. You may not need the interview you cannot get.
You do not have to get picked. You can build something someone will pay for.
If you are in that spot and want to talk it through, call me at +1 (425) 979-2282. I pick up. If I am with someone, leave a message and I will call you back the same day. Or write to john@common-sense.com. I was there, and I am glad to talk.