A company that cuts every paycheck in town is cutting its own customers' budgets. That is the buyer problem with AI replacement, and on my Tuesday LIVE, October 6, 2026, I said almost nobody is talking about it.
Every paycheck is somebody else's sales
A paycheck does not sit still. It leaves the building on Friday and gets spent by Saturday: groceries, the car payment, the mortgage, a pizza, the kid's soccer fees, a haircut. One company's payroll is a hundred other businesses' revenue. That is not a theory. That is how money moves through a town like Santa Clarita every single week.
So when an owner looks at AI and sees a salary he can delete, he is reading half the page. The salary leaves his expense column. It also leaves the pool of money that buys his product, and the product down the street, and the house across the way. One company doing it barely registers. Thousands of companies doing it in the same stretch is a different animal.
I said it on the LIVE. I still mean it today.
"Who's going to buy your stuff? I still think that's prevalent. I don't see people talking about it."
Shrink it down to one cul-de-sac with ten houses. Say two earners on that street lose their jobs to AI in the same year. Those two households stop eating out, stop paying the pool guy and stop calling the plumber for anything short of an emergency.
The pool guy, the plumber and the pizza place each lose a little. None of them got replaced by AI. They lost customers to it.
The reason nobody talks about it is simple. Every spreadsheet is built for one company. There is no line on a quarterly report for "customers we laid off." The savings show up right away in one place. The lost spending shows up later, spread across a thousand other businesses, where nobody connects it back to the decision that caused it.
I first worked through this idea in an earlier piece, AI Replaces Your Workers. Then Who Buys Your Product? This one goes further, because on Tuesday night I told a story about a guy named John, and I offered a fix.
Turn the unemployment dial in your head
Here is the thought experiment I ran on air. Picture AI displacing 5% of the workforce, which is already a problem. Now make it 10, then 15, then 16 or 17. Somewhere on that dial, a bad year turns into a horrible place to be.
The damage does not stop with the people who lose their jobs. The ones still working watch their neighbors get cut and start holding their wallets a little tighter. They skip the remodel, push off the new car and stop going out on Friday. Fear spends less than confidence does, and that is my read after decades of watching people make money decisions.
This was on my mind because of a headline that week. Executives from OpenAI, Meta, Google and others testified at a New York City Council hearing on AI risks, and an AI researcher there warned about racing to build and grow our own adversary. I put it this way on the LIVE: "A researcher told New York City, we are racing to build our own adversary. That's a big problem."
Most people hear "adversary" and picture robots. I hear something more boring and more likely. A tool that works so well that the people running it forget who their customers are. You do not need a rogue machine to hurt an economy. You just need a lot of smart people making the same short-term move at once.
John, Clyde and the partnership pitch
This is a story I made up on the LIVE, but you will recognize every piece of it. John has worked at his company for 20 years. He does Zoom calls with clients.
He sells product, or support, or whatever the company does. It does not matter. What matters is that John is good at it and the clients know him.
One day HR calls John in. The powers that be want to help you, John. We want you to work with AI. A partnership of sorts. It might reduce your workload.
It might make your evaluations look better, because now you are showing everybody what a team player you are. Maybe there is a bonus in it. Maybe you skip your review at the end of the month. They tease you into it, and John says yes all day long.
So John gets his AI. It might be an enterprise version of one of the big public models, or it might be the company's own model running on its own hardware, trained and dialed up to work alongside John. In my story the AI is named Clyde, after the monkey in Every Which Way But Loose. Am I dating myself? Probably.
Clyde is great. Clyde is curious and attentive. Clyde talks the way John talks and likes the same teams John likes. Within a few weeks, Clyde knows exactly what John does.
John's workload gets lighter. He barely has to think. A couple of months in, he gets a pat on the back and maybe a little bump. The company kept its word, and John is a happy man.
The meeting John was not in
Three or four months later, the boss sits down with Clyde. Not with John. With Clyde. What do you think of John? John's a good human, Clyde says.
Do you understand what John does? Yes. Lay it out for me. And Clyde lays out John's whole job, chapter and verse.
Then the boss asks the real question. If John had to move on tomorrow, could you do his job? Clyde says yes. How well? I gave Clyde this answer on the LIVE:
"Well, I could probably up his productivity by 100% if he wasn't there to slow me down."
And Clyde adds the kicker. Don't forget, I'm not here eight hours a day. I'm here 24 hours a day, 7 days a week, 365 days a year. The boss likes that a lot.
So they call John in. We are downsizing. Thank you for everything. We know you are three months away from being vested, but we are going to have to let you go.
They shut down his office and rent it out. Clyde lives in the cloud now and does John's job without a hiccup. And Clyde can show up on a Zoom call looking just like a person. I said on air that as of that night, AI already looks that good on a video call.
Notice that nobody in the story is a cartoon villain. HR believed the pitch. The boss answers to numbers. Clyde answered the question it was asked.
That is what makes it uncomfortable. Every step was reasonable, and John still ended up on the curb with his vesting date in the rearview mirror. My take on where the responsibility lands is in The Machine Didn't Sign The Layoff.
Do the math on Clyde's offer and it points the other way
On the LIVE I fumbled the arithmetic, so let me lay it out clean. A day has three eight-hour shifts. John covers one of them.
Clyde says it can double John's output inside that one shift. Now run Clyde across all three shifts at that same pace and you get six times what John produced alone. That is the 600% I was reaching for on air.
Now look at a week. If John works a standard 40-hour week, he covers 40 of the 168 hours in it. That is less than a quarter of the clock. The other three quarters are nights, weekends and holidays, the hours when John's company has been closed or sending calls to voicemail for 20 years.
Here is what the boss in my story missed. Firing John and handing his shift to Clyde captures one shift of improvement. Keeping John and adding Clyde captures all three shifts plus John: the after-hours call from a frustrated client, the Saturday lead who was ready to buy, the holiday emergency nobody answered. That is work the company was never doing at all.
Said another way, the 24/7 argument is an argument to add capacity, not to cut people. If Clyde really is that good, the company should be chasing the business it has been missing, not shaving one salary off the business it already has. I ran a version of this math earlier in the year in AI Job Loss Does Not Have To Happen, and Tuesday night reminded me how few owners are running it.
My fix: keep John, then hire John 2
I promised a solution on the LIVE, so here it is, and it is mine. It is a proposal, not a law and not a prediction.
"Don't get rid of John, for God's sakes. Keep John."
Then go one step further. Since AI is taking over so much of the work, bring on one more person, a John 2. Pay the salary. Pay the benefits. The whole nine yards.
Pick somebody who got knocked out of the workforce and needs income. What will John 2 do? I said as much on air: we do not know the answer to that yet. But the company that just multiplied its output can afford to find out.
I said the same thing two days earlier, on my Sunday LIVE, before I had the John story in my head. "Why don't you just keep the people, right? Why don't you keep them? Don't get rid of them."
AI gets the two extra shifts. AI gets the holidays and the weekends. AI gets the after-hours support. The people stay paid, and they do not lose their homes.
I also admitted the weak spot. It comes down to money. An owner who keeps John is choosing a smaller margin today.
That is why I said there probably has to be some kind of rule, some decision made at a higher level, because left alone, plenty of companies will take the margin. I am not thrilled about trusting politicians to write that rule. I just do not see the market writing it on its own.
What keeping John actually buys you
If you are the owner, keeping John is a real cost. Payroll, benefits, a desk or a laptop, the whole thing. Somebody above you is going to ask why you are paying a human for work the machine can do. You need a better answer than "because Connor said so."
Here is the answer. John is the person who knows which client is about to leave before the numbers show it. John is the one who can tell when an answer is technically right and practically wrong.
John is the escalation path when Clyde gets stuck on a case it has never seen. John is the person who checks the machine's work before a customer ever sees it. Clyde can be brilliant and still be confidently wrong, and somebody has to catch that.
There is also the plain risk of dependence. If the tool changes its price, changes its rules, goes down for a week or disappears, the company with John still knows how to do the job. The company without John has a blank screen and a phone ringing. I wrote about that exact risk for myself in Keep the Skill Under the AI.
John 2 is the harder sell, and I know it. The best version I can see is this: as Clyde starts covering nights and weekends, the volume goes up, and more volume means more escalations, more relationships and more output to check. John 2 grows into that work. If the business never grows, John 2 is pure cost, and I will not pretend otherwise.
Five people, one box of chips, and the high-rise down the street
Now flip it around. On the LIVE I pulled four people out of the chat and said, let's start our own company. We do not rent an office. We use my apartment.
I scrounge up enough money for one of those powerful AI boxes loaded with the fancy chips, and we build a company that does exactly what John's company does, out of that tall building with all those people in it. We charge half. We get better reviews. We have zero human workforce besides the five of us.
In that version of the story, John's company folds, because it cannot compete with an all-AI shop with agents that talk to customers on Zoom and look like real people. That is the pressure every incumbent is going to feel, whether the owner keeps John or not. I am not going to tell you keeping people is free when a competitor with no payroll is coming for your price.
But that small shop has a weak spot too. Right now a lot of AI customer service frustrates people. You know the drill. You figure out it is a bot, it cannot get outside its guidelines, and you start saying "operator, operator" into the phone and nobody comes.
The 57-year-old in me gets ticked off fast. Once these systems can really solve problems and hand off on their own, people will sign on the dotted line and move on. Until then, a human who can fix it is a selling point.
There is also the question of telling people it is AI. On air I said I do not know whether customers would care if the machine did a great job. On my own site, SantaClaritaOpenHouses.com, my assistant Clara tells people she is an AI first thing.
That is the standard I chose. If you are the incumbent, your edge over the five-person shop is the people who already know your customers. Use them as the quality layer the cheap competitor cannot copy yet.
How a layoff walks into a neighborhood
This is my lane. I have been in real estate since the late 1990s, and I have watched what happens to a neighborhood when paychecks stop. Here is how I put it on Tuesday night:
"Once you get above a certain amount of people unemployed in any part of the world, then you have foreclosures begin, you have property prices start to dwindle and come down."
The chain is not complicated. The paycheck stops. Savings carry the house for a while. Then a mortgage payment gets missed, then another.
Some owners sell in a hurry, some go short sale, some end up in foreclosure. Meanwhile the HOA dues stop coming in, but the association still owes for insurance, landscaping and its reserves. That shortfall lands on the neighbors who are still paying, through higher dues or special assessments. More homes hit the market at the same time fewer people can buy, and prices soften.
That is my reasoning about how displacement would flow into housing. It is not a forecast for Santa Clarita. As of the October 6 pull from SantaClaritaOpenHouses.com, the valley had 841 active listings and 152 homes in escrow, down 24 from September 29.
Distress was a small slice: 30 short sales, 3 bank-owned and 7 pre-foreclosure, about 5% of what was for sale. One week is not a trend. The full breakdown is in Santa Clarita Escrows Dropped 24.
I also asked a darker question out loud on the LIVE. Could somebody want all this to come down, so they can buy low and sell high when it comes back? I do not know. It was a question, not a claim, and I am leaving it as one.
One more thing. I am not your lawyer, your economist or your financial advisor, and none of this is financial advice. If your job or your house is on the line, talk to the professionals who can see your actual numbers.
What a smart owner does this quarter
If you run a business and AI is already on your desk, the John story gives you a better playbook than the one HR used. Here is where I would start, this quarter, before anybody gets a meeting invite they did not see coming.
- Redeploy before you replace. Map what John does. Hand Clyde the repetitive work and move John toward judgment, relationships and the new after-hours business Clyde can now catch.
- Run the 24/7 math on revenue first. Before you count salary savings, count the calls, leads and orders you have been missing at night, on weekends and on holidays.
- Measure what customers feel, not just what you save. Track complaints, repeat customers, review scores and how often the AI has to hand a customer off.
- Keep a human escalation path that a customer can actually reach. If people are yelling "operator" and nobody comes, you are training them to leave.
- Be upfront with your staff about what the AI is for. The John story turns ugly because the pitch hid the plan. Tell people what you are measuring and what happens to their role.
None of that requires you to swear off AI. I use it every day. It does heavy lifting in my own business that six months ago I did by hand. The point is to use it to grow the pie your customers buy from, not to shrink the paychecks they buy with. If you want help mapping one real process this way, start at BookWithHonor.com.
If you are John, start here
If you are the employee sitting across from the HR pitch, do not refuse the tool. John's mistake was not using Clyde. His mistake was letting Clyde become the only one in the building who could explain his job. Be the person who runs the AI, checks the AI and catches the AI when it is wrong.
Start keeping a simple log of your judgment calls. Every time the machine gives an answer you had to fix, write down what it said, what you changed and why. Every time a client called you instead of the bot, note it. That log is proof of value that never shows up on a productivity dashboard, and it is the thing you hand your boss when somebody asks what John actually does.
Keep the underlying skill sharp, too. If the tool does all your thinking for a year, you will be the person who cannot do the job when the tool changes. And know your own timeline. If there is a vesting date, a bonus date or a contract date in your future, know it, and ask direct questions early instead of three months too late.
If you already got the call, the same advice applies to the next job. The skills that made you good, judgment, relationships and the gut sense that something is off, are exactly what a company running AI around the clock needs a person for. Lead with that when you sell yourself. Do not promise to outwork the machine. Promise to run it and to catch it.
Last thing. If your company is rolling out AI and nobody has said what happens to the people, ask. Politely, on the record, early. The best time to make the case for keeping John is before the boss sits down with Clyde.
Keep reading
- AI Job Loss Does Not Have To Happen: the longer version of the 24-hour math and why replacement leaves money on the table.
- AI Won, But Who Pays The Gap?: what it would take to carry people through the transition.
- The Machine Didn't Sign The Layoff: why a person, not the software, owns every cut.
- Keep the Skill Under the AI: how to stay capable if the tool you lean on goes away.
- Santa Clarita Escrows Dropped 24: this week's local housing numbers, for homeowners watching the market.