Skip to Content

Are AI agents going to take my people's jobs?

An honest answer from an owner who uses them

Short answer: Some jobs will change. A few may go away over time, mostly through fewer new hires rather than layoffs. The way I use AI agents, my people move from doing the busywork to checking it and handling what the agent can't. You owe your team that answer straight, before anything starts.

I get this question from almost every owner I talk to. Usually it comes up second, right after "does it actually work?" And it's usually not really about the owner. It's about the office manager who's been there forever, or the bookkeeper who knows where every receipt is.

So here's my honest take.

Which jobs actually change

The work I'd hand to agents first is the repetitive stuff. Inbox triage. Data entry. Chasing follow-ups. Building reports. First-pass bookkeeping. Prospect lists.

If that's most of somebody's day, their job is going to look different in a year. I'm not going to tell you otherwise.

What it looks like at my company: the AR agent drafts invoices on the billing calendar and tracks what's overdue. The person who used to chase every invoice now checks the drafts and handles the ones that need a phone call. That's a real change in the job. It's also a better use of somebody who knows the clients.

The pattern is the same across most of what we've set up. The agent drafts. A person decides. The job moves from doing to checking, and from typing to judgment.

How the work changes: inbox triage, data entry, chasing follow-ups, building reports, first-pass bookkeeping and prospect lists go to agents. People keep checking the drafts, handling what the agent can't, the calls that need a person, approving anything a customer sees, and judgment.

What I did at Zelifcam

I didn't let anyone go to build this. I use it so the people I have spend their day on clients instead of copy and paste.

I'll be straight about the other side too. When you have agents doing the busywork, you think harder before you post a job for more busywork. That's the honest version. If somebody tells you AI won't change hiring at all, they're selling something.

Why "human in the loop" isn't just a slogan

You'll hear "human-in-the-loop" a lot right now. At my company it means something specific: anything new with my name on it waits for my OK on the exact text. An email to a client, a post, a proposal. An agent can draft it. A person reads the actual words and says yes.

That's not only about avoiding mistakes, though agents do make them. It's about accountability. A customer should never get something from your company that no person at your company read. And somebody has to own that check. That's a job, and it's a job for a person who knows your business.

How to talk to your team about it

This is the part most owners skip, and it's the part that matters most. Your people are going to hear about the agents one way or another. They should hear it from you, first, before anything starts.

Here's what I'd say, in your own words:

  • "These agents draft. You decide."
  • "Your job is moving from doing the busywork to checking it and handling what the agent can't."
  • "A person approves anything a customer sees. That person is one of you."
  • "Nothing here scores you or watches how you work. It's looking for client questions that slipped, not grading anybody."
What to tell your team before anything starts: these agents draft and you decide; your job moves from doing the busywork to checking it; a person approves anything a customer sees; nothing here scores you or watches how you work.

If you're willing to go further, some owners make a simple promise: nobody loses their job because of this project for a set period, and any time it frees up goes to other work. You don't have to make that promise. If you do make it, keep it.

The rules we hold ourselves to

When we set agents up for other companies, a few rules come with us. We tell your staff the truth about what the agents do. A person approves anything a customer sees. We don't build anything that secretly monitors or scores employees. And we won't run a project whose real goal is a layoff unless the owner tells the team that first.

So, should you be worried?

If you're an owner, I'd be more worried about the busywork eating your best people's week than about the agents. If you're an employee reading this, the safest place to be is the person who knows the work well enough to check it.

If you want to see how this works in practice, here's how I run my company with a team of AI agents, and here's what we set up for other owners.

See how our AI ops team is set up

Related: What our software factory actually does

Brandon Howard is the CEO of Zelifcam, a custom software, automation and Odoo partner in San Antonio, Texas.

How I run my company with a team of AI agents (and still approve everything new)