lab-reports

Will AI Take Your Job? I'm Building the Thing That Might.

An agency founder building 12 AI agents shares a candid look at jobs, risk, and where AI replaces tasks vs. people—and what to do this week.

By Ben Kalkman · 9 min read · AI and Jobs AI Strategy Automation Agencies Home Services Workforce AI Adoption Career Development

On October 4, 60 Minutes ran a segment asking the question I hear on almost every podcast I'm on: will AI take my job?

I've had the same answer for years. No. You probably won't lose your job to AI. You'll lose it to someone who knows AI better than you. I didn't coin that line. I've just said it enough times that it feels like mine.

But watching the segment, I had to sit with something uncomfortable. Right now, in one of my companies, I'm building 12 AI agents designed to do work that people would normally do.

So my easy answer deserves a harder look.

Three people, three answers

The segment set up a real debate, and I think each side is partly right.

The optimist. Brendan Foody, the CEO of Mercor, runs a company that has gathered more than 100,000 freelance experts to teach AI models what they know, from finance to poetry. He sees this as a natural evolution: AI takes the dull, repetitive parts of the job and frees people for more meaningful work. (CBS News)

The alarm. Clara Shih, who led major AI divisions at Salesforce and Meta, quit her job in January to start a nonprofit for young workers. Her point: the work we used to hand a 22-year-old (market research, the first draft of a memo) is exactly what AI now does well. (CBS News)

The economist. Nobel laureate Daron Acemoglu argued that the first phase of the Industrial Revolution played out over 80 years, while this shift is happening in one or two, across many industries at once. In a worst-case scenario where nothing changes, he said unemployment could triple in the next decade. (CBS News)

And the fear is real. 60 Minutes cited a Pew Research study finding that nearly three-fourths of Americans now worry AI is coming for jobs (Pew Research Center). Challenger, Gray & Christmas reported that AI was the leading reason companies gave for job cuts for five straight months through July 2026 (Challenger, Gray & Christmas). Worth noting: Challenger also warns that some companies blame AI for cuts that are really about cost, without AI ready to fill those roles (HCAMag).

I'm not watching this debate from the outside. I'm running both experiments at once.

Two companies, two experiments

Rocket Media has served home services companies for 23 years. Inside it, we run something we call Rocket AI Labs. My belief is simple: when it comes to AI, your organization is only as strong as its weakest link. The more knowledge you can get out of people's heads and into a shared system, the stronger everyone gets.

We're in our annual business review season right now, and we're treating the whole process as a Labs experiment: how do we rebuild annual planning for our clients with AI at the center?

I also hire differently now. I look for people who already think in AI, who have a process for using it, and who can show me where they'd make an impact. And I'll be honest: we are building bots to handle some functions we no longer need people to do. But we want people overseeing them.

Digital Ignitor is the other experiment. It's a newer company, and I'm using it to see how far I can push AI as the one doing the work. We're building 12 AI agents. I plan to hire one or two people to sit in the middle as the truth tellers. Everything else is automation.

I put it under a different brand on purpose. I didn't want the experiment to touch what Rocket Media has built over two decades. The plan is to learn everything I can, document it the whole way, and bring the best of it back into Rocket Media.

Side-by-side comparison of Rocket Media, where AI makes people stronger, and Digital Ignitor, where 12 AI agents do the work with one to two humans overseeing, with a dashed arrow showing lessons flowing back and a question mark over what it means for the team.

Which brings me to the part I haven't figured out.

The part I haven't figured out

AI has already cost us someone.

One of the people on our content team resigned and left the industry entirely. Not for another agency. A completely different field. She was frustrated with where AI was taking the craft she'd built her career on.

Now I'm seeing the same thing start on our web team. No resignations. But a lot of pushback and real strain, as AI slowly creeps into work they spent years getting good at. It happened fast.

And here's the question I keep coming back to. If Digital Ignitor proves that 12 agents and two humans can do the work, what happens when I "bring the best back" into Rocket Media?

I don't have a clean answer. I know what I hope: that AI takes the repetitive parts and our people move up to the work only humans can do. But I'd be lying if I told you I know exactly what that looks like for every person on that team.

Clara Shih said it best on the segment: we have to be honest with people about the risks. So that's me being honest.

AI answers the phone. It doesn't turn the wrench.

Here's what I'm seeing across our home services clients, the HVAC, plumbing, electrical, and solar companies we serve every day.

The shift is happening at the front desk. Some are using AI receptionists. Many use AI for scheduling, auto responses, and follow-up surveys. Some use it for training.

Out in the field, it looks different. Our clients use a booking and field platform where a tech can snap a photo of a unit's serial number and pull up its entire service history. They can photograph a problem and get help diagnosing it, along with recommended parts and part numbers. And they can tap into a shared knowledge base where teammates post how they solved the same issue last month.

That's not a future scenario. It's what our clients' techs are doing today.

But nobody's AI is installing an air conditioner. Nobody's AI is up on a roof fixing a solar panel. AI is making the tech faster and smarter. It isn't replacing the tech.

A spectrum from more AI to more human across a home services company: AI handles the front desk, AI assists the field tech, and installing equipment and turning the wrench stays human only.

Pickleball doesn't scale

I believe there's a difference between efficient and good.

Every business chases efficiency, mine included. Efficiency can mean better margins and a better product for clients. AI is the best tool I've ever had for finding it.

But something can be efficient and not good. And something can be good and look inefficient.

A two-by-two grid of efficient versus good. A pickleball review with a client is highlighted as good but looks inefficient; AI-built review reports are efficient and good; a bot replacing every human touch is efficient but not good; busywork is neither.

One of our local clients prefers to meet in person for our review calls. Somewhere along the way, those reviews started including pickleball. AI is in the room: it helps build the reports we review together. But it can't play the match. It can't read body language, hear the shift in someone's voice when a number worries them, or build the kind of trust that comes from breaking bread together.

Kim Herlihy, a trial lawyer interviewed for the segment, put it this way: "You call your lawyer because you want to look in their eye." (CBS News)

On paper, an afternoon of pickleball is inefficient. In reality, I'd argue it's one of the most efficient ways I can spend my time. That's how relationships get secured and accounts grow.

The interns and the second grader

We've run an intern program for years, and we let it slip the last couple of seasons. It's coming back next summer. We'll train our interns to use AI the right way on the work you'd normally hand a younger team member.

But I have to admit the tension Shih raised. Judgment usually comes from doing the grunt work badly first. If AI does the reps, how does a 21-year-old learn to spot when AI is wrong? I don't know yet. That's part of what next summer is testing.

At home, I see the whole range. Two of my daughters are in college, both pursuing the medical field, work that depends on high-touch human interaction. My son is in the trades. And six more kids are still coming up behind them. My youngest is in second grade.

"Worry" isn't quite the right word. It's the unknown. No one has a crystal ball for what AI looks like in 5 or 10 years, and a lot will change before my second grader graduates high school.

What does frustrate me is how many schools, especially colleges, still treat AI like something to catch rather than something to teach. I think it should be woven into nearly everything students do, because that's the world they're walking into.

So, will AI take your job?

Here's my updated answer.

I'm still an optimist. My opinion, not a forecast: I think we'll end up with more jobs, just different ones. New roles we haven't named yet. And I think hands-on trade skills will climb toward the top of the list.

But Acemoglu has a point that my favorite line doesn't fully answer. If one person who knows AI well can do the work of three, two people still lose their jobs. And the timing matters. Housing is close to out of reach for a lot of people in their twenties, and everything costs more. A job loss right now lands harder than it would have a decade ago.

So: more jobs eventually, I believe. Real pain in between, I know. Both are true.

Start here this week

If you're a business owner worried about your team, or a parent worried about your kid, the first step is the same. You can't guide anyone through this if you don't understand it yourself. You don't need to be an expert. You need to understand enough to help.

Here's where I start everyone: build your master prompt.

For AI to be useful, it needs to know more than your personality. It needs to understand how you work, what you know, how you make decisions, and everything in between. A concise master prompt you can hand to any AI tool is the foundation. Everything else gets built on top of it.

Write it this week. Then use it, and watch how different the answers get.

Want help building yours? Building a master prompt is the first step of a full curriculum we walk business owners through at Digital Ignitor. If you want a guide for what comes next, for yourself or your team, let's talk.

Sources