Five Years Out From Not Working Anymore

Listen to the episode with Shahid Durrani

The short version

• My crazy prediction: we are five years out of not working anymore, and about five years from every house having one or two robots in it.
• The signal most people miss is how good AI already is. They chat with it instead of deploying it like an employee.
• We put AI agents on the phones in 100 plus physical therapy locations over the last year. More money, bigger staff, more doctors, more patients, no turnover.
• Give your agents permission to actually solve things, including refunds. A feedback loop beats an angry customer.
• This interview exists because Luna, one of my agents, listens to podcasts every day and pitches the hosts.
• Manufacturing, food service, medical and transportation get disrupted first. Trades and highly creative work hold on the longest.

The prediction that sounds crazy

Shahid asked if I had any predictions that sound crazy today. Mine is simple: I think we are five years out of not working anymore. That does not mean nothing gets built. I think there is going to be room to be an entrepreneur, and room to own the machines or own the intellectual property that is powering the machines. I just do not think we are going to work in the same way that we work today.

I am living a bit of the future here in Austin because we are in Elon's playground. When I need to go downtown, I hop in a robotaxi. It is $5, I do not have to talk to anybody, I get in the car and it knows who I am. Tesla cars here in Austin are delivering themselves, so if you order a car it can leave the assembly line and drive to your house and park out in front of your house. That is one of the reasons we chose to live here.

I also think we are probably five years away from every house having at least one or two robots in it that cook and clean. I do not think we are going to shop at grocery stores the same way. It arrives at the house and the robot goes to the curb and picks it up out of the truck. My mom was a school teacher and she worked until she got disabled, on social security, when she had cancer. I do not want to work my life away, and I am glad future generations may not have to.

The part that scares me is the adoption curve. You are going to have people who are overly AI and embrace everything, and people who are anti AI and fight every aspect of it, and for the next three to five years we are going to get in some battles. Right now the easiest way for me to create an illustration is with AI, and then Instagram puts a little tag next to it saying it was created by AI. I could hire a graphic designer to make the same thing and there would be no tag. Beyond that, I am honest that we do not know what we do not know yet.

I mean, I think we're five years out of not working anymore.

The signal most people are missing

In the circles I run in, people still focus on the chat aspect of AI. I can prompt it, I can chat with it. They are not setting it on its own to go do its job. They use AI when it is convenient to answer a question instead of going to Google, and then they decide that is what AI is. The signal they are missing is that it is better now.

We deployed AI agents in 100 plus physical therapy locations over the last year, and we answer the phones. Everyone was scared at first. They said, well, what is our front desk person going to do, we do not want to fire them, they are great. My answer was to let them love on the people when they walk in the clinic.

Think about what happens without that. You walk in as a brand new patient and say hey, I am Gary, I have an appointment at 11AM, and the front desk says hold on one second, I am on the phone. You just told the human in front of you to wait so you could talk to somebody who is not there. And when you are back getting treated, it is phone ringing and phone ringing and phone ringing. Let AI handle that and take care of the people who showed up.

When we did it, it totally rewired the way the clinics work. More money, bigger staff, more doctors, more patients, everything better, because we let AI do the job. And we did not have turnover. We have a consistent employee that works twenty four hours a day, seven days a week doing exactly what we need it to do.

So they're not deploying AI like they would an employee.

Give the agent permission to actually help

A lot of people put AI in a little box and they do not empower it to serve the customer. They treat it like it is there to answer questions, like an FAQ agent, and anything real goes to humans. That is not the interesting version. Give AI permission to give refunds. Give it permission to give extra credits if you run on credits. You can put guardrails around it, but the guardrails should not be so tight that the agent cannot solve anything.

There is that saying that it is easier to ask for forgiveness than permission. I would rather my agent give someone a little bit extra, catch it in a feedback loop and fix it so we do not do that again, than have AI make someone mad because it cannot even solve their problem. The companies I feel best about are the ones where I forgot to cancel a subscription, got the receipt, emailed them, and they wrote back and said Gary, no problem, we refunded you. So set a rule: within forty eight hours of a bill, give the refund and tell them it is settled. They did not want to be a customer that month anyway, and they may come back. I have done that with Restream over and over because they give me great service.

We run this on ourselves. We have a community of 600 and some people, and support tickets come in like they do anywhere. We built a support engine, an AI agent reads the ticket, fixes the problem, and I never have to get involved. I hate support, and I hate giving slow support even more. If you have a problem on a Sunday, that is when I want it solved, not Monday when I get around to the queue.

So if you have a problem, I want you to get it solved at 7AM on Sunday morning when you want your problem solved.

Luna booked this conversation

AI takes the busy work away, and I mean the busy work I would never do myself. I have an AI agent named Luna, and her job every single day is to go out and listen to podcasts, find shows I would make a great guest for, and reach out to the host to see if it is a fit. I had three interviews that week, all booked by Luna. Years ago my wife did this for me. She hated it, it was tedious, and she was in an angry mood every day because she had to keep reaching out to people and listening to shows.

Luna can listen to the shows. She can understand what a host talks about, decide if there is a real match, and write a pitch that explains how we align. That is how Shahid and I ended up talking at all. The part he liked most about the pitch was the disclaimer, because we say up front that an agent did the outreach. Transparency is the whole thing. If you are going to let an agent speak for you, say so.

Two hours of school a day

About a week and a half before we recorded, I met two people who work for a school called Alpha School. Their high school tuition is $150,000 a year, so it is expensive, it is a private school. Their guarantee is that when you graduate high school you have a startup with a valuation of a million dollars or more, and if you do not, they refund all your tuition money. And they only have education for two hours a day.

Two hours works because of AI guided models. You are going to learn differently than I do, and every listener learns differently than either of us. If we tailor education to the human, we can teach the same concept in different ways and speed track the amount of time you need to spend in instruction. We do not need a babysitter while parents go to work anymore. We need to educate and enrich kids so they can go build and grow.

I saw a sports school here in Texas doing the same thing. Two hours of education, then basketball or whatever their high school sport is, with world class trainers coming in to work them out in the gyms and train them to go be a college or professional athlete. When my teachers yelled at me in the nineties, they said I would not have a calculator in my pocket every day and I had to learn the math. Thirty years later I have a phone with every model on it and access to anything in the world at a second's notice.

What gets disrupted, and what survives

Shahid asked me for the industries that change completely. Honestly I think every industry will, but here is the order I see it happening.

• Manufacturing, and it is really close right now. Robotics on the assembly line instead of humans for cars, televisions, furniture. Those are centralized places, so they go fast.
• Food service. Why do we need a human to drop the fries down. Smart robotics handle that, and the drive through order gets taken by AI, which means a smoother and more predictable experience than getting ketchup on a burger you did not want ketchup on.
• Medical. I think we eliminate a lot of doctors and nurses, because a doctor who diagnoses faster with AI sees more patients. There is a Chinese hospital right now that is 100% AI and robotics with no human oversight at all. And if I am doing telemedicine, I do not know why I need a human on the other end. AI can read an x ray the moment it is done instead of waiting for someone to get to it.
• Transportation. I cannot think of a single reason I need an airline pilot flying me anymore, and most of the time they are on autopilot anyway. Elon says in ten years, 90% of all travel will be autonomous driven.

The job I think is safe is the job nobody really wanted, which is blue collar work. Plumbers, electricians, those trades are safe for a long time, because it is going to take a lot to get a robot that can dig under your house and fix your plumbing. My son has an AI agency he is building, but if he told me he wanted to go to school, I would tell him to go be a plumber or an electrician. Middle management is the other side of that, because we do not have to manage robots the way we manage humans. What holds is highly creative work and the people who lead a fleet or a team of agents, and I do not think even that is a ten year job.

The economics are what force the issue. AI makes everything much cheaper because the cost of labor goes down, so you can manufacture a TV cheaper and a car cheaper. The businesses that do not adopt still have to pay human capital, and human capital keeps rising because we have a high self worth. I need a raise, can you give me a raise. For a while, not adopting is a novelty, we still have human servers and a human this and a human that. Eventually they wear out, because my 18 year old and my 14 year old would more than likely prefer to deal with AI than a human.

The AI doesn't ask for a raise. It just does the job.

Attention is shrinking, and that is not the whole story

I do think attention spans are getting shorter and we are compressing our learning. I am biased here, because I was very big on Clubhouse. I grew about 90,000 followers there in a couple of months in the business rooms, and I wrote a book called Clubhouse Creator that was published by Hay House. What I learned from audio is that there are different modalities to connect in. With video there is a judgment piece, I see you and you see me and people judge what they see. Strip it to audio and you hear what someone says instead.

So yes, attention gets smaller, but we are still yearning for connection. I think we end up having longer conversations with the people we align with, and it is okay not to align with everybody. Build a diverse group anyway. Have conversations with people who believe the exact opposite of you, and be open minded that what you think may not be right.

That is also where AI has a real weakness. AI is very biased to give us a yes rather than to challenge us on our thought process. Tell a model you have a great idea and it will tell you it is a million dollar business and you can go crush it. A friend will say okay Gary, but I know this part of you, how is this going to play out with what you are already doing. That nuance is the human aspect we need to keep in play.

For focus, the trick is intention with a goal in mind. If you sit down not knowing what you are looking for, you have no attention to give anything. I am a micro content person, I will watch a clip before I listen to a whole recording, and then I go dig into the underlying knowledge behind it. When I got interested in peptides I pulled a lot of science and research in a short period of time, and now I have an agent that keeps me in the loop on new updates. I go to Grok first for information because it has access to the x data instantly, then I build in Claude and use ChatGPT for images and codex work.

Boring routines and the people around you

When Shahid asked about the habits that separate top performers, my answer was routine. People will tell you to wake up early, and I do, but the real thing is having a routine at all. The more boring you become on the things that do not matter, the better you perform. It does not matter what shirt you wear or what your hairdo looks like. Nick Saban ate a chicken salad every day for lunch because it was one less decision he had to make while he was making every decision for Alabama football.

Minimize your decisions. Go to bed at the same time as much as you can, wake up at the same time, and get consistent with yourself. Motivation is going to fade, so you need discipline and a bigger purpose behind it. I shave my head and shower every morning, not because I am on video that day, but because that is the routine.

Do not try to mimic one person, though. Learn from 50 people or 100 people, pull the pieces you can implement, and build your own routine. Commit to something for a couple of weeks and judge yourself honestly. Do I have brain fog, am I sleepier, am I more creative. If it does not move you toward what you are trying to accomplish, drop it.

And environment matters more than you will ever realize. If your friend group stays up late and parties every night, you slip into it. A lot of people figure that out, exit the group, and then sit over there by themselves trying to build a new business or a new life alone. That is the biggest takeaway I can give anyone: you cannot do life alone. Surround yourself with people who are on the same journey you are trying to be on.

And I think the more boring you become, the higher performer you are.

Questions people ask about this

What is Gary Henderson's prediction about work?

That we are five years out of not working anymore. He still sees room to be an entrepreneur and room to own the machines or the intellectual property powering them, but he does not think we will work the way we work today. He also thinks we are about five years from every house having one or two robots in it that cook and clean.

How are AI agents being used in physical therapy clinics?

Gary's team deployed AI agents in 100 plus physical therapy locations over the last year to answer the phones. The front desk staff stayed and focused on patients walking in instead of putting people on hold. The result was more money, bigger staff, more doctors, more patients, no turnover, and an employee that works twenty four hours a day, seven days a week.

Which industries does Gary think AI disrupts first?

Manufacturing with robotics on the assembly line, food service including drive through ordering and food prep, medical including telemedicine and reading scans, and transportation. He also expects a lot of middle management to go, because you do not manage robots the way you manage humans.

Which jobs does Gary think are safest?

Blue collar trades. Plumbers and electricians are safe for a long time because it will take a lot to get a robot that can dig under your house and fix your plumbing. Highly creative roles and the people who manage fleets of AI agents also hold, though he does not think agent management stays a human job for ten years.

How did Gary get booked on this podcast?

An AI agent named Luna. Her job every day is to listen to podcasts, find shows Gary would be a good guest for, and reach out to the host with a pitch explaining the alignment. He had three interviews that week, all booked by Luna, and the pitch included a disclaimer that an agent did the outreach.

How much authority should you give a support agent?

More than most people do. Gary's view is that treating AI like an FAQ bot wastes it. Give it permission to issue refunds and extra credits with guardrails, for example a refund inside forty eight hours of a bill. He would rather catch a slightly generous agent in a feedback loop than have a customer left with an unsolved problem.

Watch or listen to the whole conversation

The full episode is on Super Entrepreneurs Podcast, hosted by Shahid Durrani.