She Was About to Hire a VA for $13/Hour. Then She Saw AI Do It for 20 Cents.

She was about to hire a VA for $13 an hour.

Then she came to my AI workshop. Two days later, we got on a discovery call. First thing she said: “When I saw Lindy schedule a meeting for 6 cents… I told the VA I’m not gonna hire anybody right now.”

That sentence stuck with me. Because it wasn’t me trying to talk her out of hiring. She got there herself.

Amanda’s story

Amanda runs a CPA and bookkeeping firm in Texas. Four employees. She’s at capacity — turning away new clients because there’s no room. She wakes up at 4:30am some mornings just to get through emails before the workday starts.

She fired her assistant in March. Not because of attitude. Because after a year and a half of training — every process documented, every preference explained, every workflow laid out — the assistant couldn’t adapt to the new tools the firm was moving to. Had to let her go.

And when she left, everything went with her.

That’s the part that doesn’t get talked about enough. When a human assistant leaves, the training leaves too. All that context about how you like things handled, what tone to use, which clients get priority — gone. You’re starting from scratch with the next hire.

So when Amanda came to the workshop and saw an AI agent doing real scheduling work… she paid attention.

The 20-cent meeting

Amanda hadn’t known our discovery call was scheduled by an AI agent until I mentioned it.

Lindy had handled the whole thing. Sent the initial email, replied within 30 seconds of every response, worked over the weekend when Amanda followed up on a Saturday. When I told her it was AI, she laughed. She’d described the replies as “pretty humanistic.”

Total cost for scheduling the whole call: about 20 cents.

She’d been comparing $13/hour to that. And she’d done the math before I said a word.

The new question before every hire

A few years ago, the decision to hire an admin assistant was pretty simple: do you have enough work for them? If yes, hire. Figure out the rest later.

That math has changed.

Now there’s a step that comes first: what would this person actually spend their time on? Before you write the job description, before you post on Indeed, before you start interviewing — list the tasks.

Email management. Scheduling. Post-meeting follow-ups. Document routing. Calendar conflicts. Answering client questions that follow a predictable pattern.

Then ask: how much of that could AI handle?

Amanda put it plainly on our call: “I’m not making any hiring decisions until I have a better understanding of what AI can do.”

That’s the new filter. And I think more business owners are landing there on their own.

What AI can’t replace (yet)

This isn’t an argument against hiring. It’s an argument for sequencing.

There are things an AI agent handles well right now: anything that’s high-volume, pattern-based, and doesn’t require the kind of relationship judgment a human brings. Scheduling. Routing emails. Drafting follow-ups. Creating meeting agendas from transcripts. Sending reminders. Answering FAQs.

There are things that still need a human: complex relationship management, calls that require real empathy, judgment calls in ambiguous situations, anything that benefits from genuine trust between people.

The practical question is: what percentage of the role is the first category?

For most admin and EA-type roles in small businesses right now, the answer is somewhere between 60% and 80%.

What I call the AI-first filter

Before the next hire, run through this:

1. List every task that role would own
2. For each task, ask: is this high-volume, pattern-based, and low-judgment?
3. For everything that lands in that category, ask: is there an AI tool that already does this?
4. What’s left over after the AI handles its share — is that enough for a full hire? Or a smaller one?

This isn’t about eliminating people. Amanda has four employees she wants to keep. She wants to grow. But she’s being thoughtful about what the next dollar of headcount buys her.

That kind of thinking is what separates businesses that scale cleanly from ones that add people and complexity at the same rate.

One more thing about Amanda’s former assistant

The story about the assistant who left stuck with me for another reason.

She’d spent a year and a half building that person up. All those training sessions. All those corrections. Every preference explained. And when the person left, the knowledge went with them — not just the skills, but the specific context about how Amanda’s firm runs.

An AI agent doesn’t resign. You build the context once — the preferences, the tone, the routing rules — and it stays. It doesn’t get a better offer somewhere else. It doesn’t need to be retrained every 18 months because someone moved on.

For small business owners who’ve been through the hiring-and-losing cycle a few times, that permanence is worth something. Maybe more than the cost difference.

Amanda and I ended up working together. She enrolled in a 30-day AI implementation. Every week, we deploy one new agent. By October 1st, her firm will have AI handling email management, scheduling, calendar coordination, and post-meeting workflows.

She’ll still have four employees. But none of them will be waking up at 4:30am to get through the inbox.