Working with AI

AI should give you your evenings back, not take your team’s jobs

By Jared Theisinger, Founder & Director · Clearline AI, Melbourne · 8 min read

Read that distribution carefully, because it is the opposite of the replacement story. The tool did not make the best people redundant. It brought everyone else closer to them. The researchers found it worked by spreading the practices of the strongest performers across the team, helping newer staff move down the experience curve faster than they would have alone.

If you have ever spent months getting a new starter to the point where you trust them with the phone, that finding should mean something to you.

What AI is genuinely good at, and what it is not

The useful dividing line is not clever versus simple. It is repeatable and low-judgement versus variable and high-stakes.

Answering a call at 8pm and taking the details. Acknowledging an enquiry within seconds so the person does not ring your competitor. Sending the reminder that stops a no-show. Chasing the quote that went quiet. None of that requires judgement. All of it requires somebody to be available at a moment nobody is available.

Quoting a complex job. Reading a customer who is upset about something that is partly your fault. Deciding whether to take work on when the schedule is already tight. Training an apprentice. That is the work that needs a person, and no amount of automation changes it.

That study is the one to keep in mind, because it puts a number on the thing vendors never mention. AI is not uniformly helpful. Pointed at the wrong work it makes people measurably worse, and confidently so. Picking the right tasks matters more than picking the tool.

The replacement question, answered properly

It would be dishonest to say automation never costs anyone a job. At large scale, in businesses where a role is entirely repetitive processing, it clearly can.

But that is not the situation most Australian trades and service businesses are in. The typical operation I speak to is not overstaffed. It is a handful of people doing more than they can reasonably fit in a week, with the owner absorbing whatever is left over at night. There is no surplus to cut. The constraint is time, not headcount.

In that setting, automating the admin does not produce a redundancy. It produces capacity. The person who was spending an hour a day on reminders and callbacks is now doing the work you actually hired them for, and the growth that was always going to require another pair of hands can happen before you can afford them.

The direction it goes is a leadership decision, not a technical one. The same system that frees up five hours a week can be used to give someone their afternoons back or to justify cutting their hours. That choice sits with the owner, and it always did.

Where the hours actually come from

It is worth being concrete, because “saves you time” is the emptiest phrase in this industry. Here is a modelled week for a small operation.

Six weeks. That is not a productivity statistic, it is a fortnight of annual leave you did not take and a month of evenings you did not get.

The part nobody puts in the brochure

Most writing about business automation stops at the money. The part that matters more, in my experience, is what an always-on business does to the person running it.

It is the phone checked at 9pm because an enquiry might have come in. The Sunday afternoon spent on invoices and callbacks. The low background hum of holding a list of people you have not got back to yet. None of that appears in a P&L, and all of it accumulates.

I spent enough years in the Air Force to have watched what sustained operational tempo does to good people. Capability is not just what your team can do on a good day. It is what they can keep doing, week after week, without something giving way. A business that only functions because the owner never switches off is not a strong business. It is a fragile one with a single point of failure, and that point of failure is a person.

Automating the after-hours response is not a productivity play in that light. It is the difference between a business that can run for a decade and one that burns out its best person in three years.

Where to start, and what to keep human

If you take one thing from this, make it the sequence. Automate the work that is high volume, low judgement and time critical, in that order:

• First contact. The enquiry that arrives when nobody can take it. Phone answering, missed-call text-back, and the inbox, because these are pure timing problems with no judgement in them.
• The chase. Reminders, confirmations, and the follow-up on a quote that went quiet. Consistent, tedious, and the first thing a busy person drops.
• The forgotten list. Past customers nobody has contacted in two years, because that is work that will otherwise simply never get done.

And keep with people: the quoting, the difficult conversations, the judgement calls, the relationships. Not because a machine could not attempt them, but because the jagged frontier research suggests that is precisely where it would make you worse.

“I have never met an operator who wanted more hours in the business. They wanted fewer hours doing the parts of it that never needed them in the first place.” Jared Theisinger, Founder & Director, Clearline AI

Part of a series on where trades and service businesses lose work: what a missed call costs, why calling back does not recover it, the cheapest leads you already paid for, and the inbox is the leak you cannot see. More about why Clearline exists.

Last updated 14 August 2026