Amanda's tax manager went on leave right before tax season.
The timing couldn't have been worse. Tax season is the most demanding stretch of the year for a CPA firm. Losing a key team member at that moment would normally mean one thing: bring someone in fast.
Amanda took a different approach. She wanted to see what AI could handle before defaulting to a hire.
Tax season ended. Her clients were happy. The work got done. And Amanda was left with a question she hadn't expected to be asking: “Was the hire even necessary?”
What She Said
“I don't want to hire another person until I fully know every single thing that AI can do. What I did during tax season — I reflected on that a lot. It would not have been humanly possible without AI.”
That's not a casual observation. Amanda had been building out AI systems for her firm over several months — a custom GPT trained on her processes and tax knowledge, an email inbox manager, communication workflows. When her tax manager left, those systems were already in place.
The capacity gap she'd expected to feel badly didn't show up the way she anticipated.
She went into that tax season with nine employees. After the season, she was at four — and maintaining the same client quality she'd had before. Her outsourced team in India was so impressed by the custom GPT that her team lead asked where she'd gotten the training. The GPT was handling complex tax queries more precisely than anything they'd seen from a tool like that.
The thing Amanda was building toward wasn't fewer employees for its own sake. She was building toward a different question: what does this business actually need headcount for?
The Default That Needs Questioning
For most business owners, the response to capacity strain is automatic: hire. More demand, more people. It's the operating assumption that's been correct for most of business history.
But that assumption was built in a world where the only way to add capacity was to add people. That's no longer true.
AI can handle significant workload — not everything, and not always at the same quality a skilled person would produce, but far more than most business owners have tested. The problem is that most haven't run the test. They feel overwhelmed, they assume more people is the answer, and they hire without ever finding out what AI could have done.
The cost of that pattern is real. Every hire is a commitment — salary, benefits, management overhead, onboarding time, the cultural effect on the team. A hire that wasn't strictly necessary doesn't just cost money in the moment; it adds permanent structural overhead.
The Experiment Worth Running
Before your next hire, spend thirty days running a different experiment.
Take the role you're considering hiring for and ask: what are the actual tasks this person would do? Get specific. Not “manage client relationships” — the actual daily and weekly work: drafting proposals, answering common questions, preparing reports, updating records, scheduling, follow-up emails.
Then test each of those tasks with AI. Not a quick attempt — a real attempt. Build a proper system, load it with the relevant context, and see what it can actually handle.
Some things will work well. Some won't. But you'll end up knowing something most business owners don't: the AI capacity actually available to you for this specific function.
After that experiment, you'll be making a hiring decision from a much more informed position. Maybe the hire is still necessary and the experiment just confirmed it. Maybe the hire only needs to be part-time instead of full-time. Maybe a system and a smaller team can do what you thought required adding headcount.
Amanda didn't set out to reduce her team from nine to four. That happened because she kept asking the question before defaulting to the obvious answer. What can AI do here? What do I actually need a person for?
What This Doesn't Mean
I want to be clear about what I'm not saying.
I'm not saying AI replaces people. There are things that require human judgment, relationships, and accountability that AI genuinely can't replicate — and that set of things is larger than the most aggressive AI proponents acknowledge.
I'm not saying every hire is unnecessary. Some roles require a person. Some capacity gaps are real and the hire is the right call.
What I am saying is that the reflexive hire — the “I'm overwhelmed, time to bring someone in” move made without testing AI capacity first — is a less informed decision than it used to be. The space between “I need more capacity” and “I need to hire someone” has grown substantially.
The businesses getting the most out of this moment are the ones asking a new question before the old one. Not “who should I hire?” but “what does this business actually need a person for?”
Amanda found that question after tax season. It's worth asking before the next one.
I help service business owners and professional firms build AI systems that answer this question with evidence — so hiring decisions are made from a position of knowledge, not assumption. If you're thinking about headcount and want to run the AI capacity experiment first, reach out or check out my consulting and workshop programs.
