A Perth clinic went from 0-1 bookings to 12 a week by cutting a step
A Perth clinic owner, a UK board advisor and a Brazilian advisor to CEOs and founders on why the first move isn't hiring or deploying anything.
15 to 30 leads a week. 0 to 1 bookings.
One Perth clinic owner didn't buy an agent to fix that. He deleted a step.
Bookings hit 12 a week. Same ad spend.
No appointment setters. No agent. Just a shorter path from ad to booking page.
We asked which parts of a job AI has taken and which still need a person. Three voices in Australia, the UK and Brazil answered with a reframe: the answer comes before the tool, and it starts with whether the work is even defined. It matters now because every company is weighing a hire against an agent, and the costliest mistake happens before either decision.
The best automation was a delete key
Mitchell Kwan, Founder of MK Aesthetics (Perth, Australia)
“I sent the ads straight to a booking page instead. Leads fell to 1 to 2 a week. Bookings rose to 12 a week on the same spend, with no appointment setters and no agent.”
“The expensive mistake is automating a broken step. An agent chasing those 15 to 30 leads would only have made the wrong process run faster.”
Fewer leads, more bookings. The fix was subtraction, not software, and the ad budget never moved.
Wrong question, says the board advisor
Paul Wakefield, CPD-certified Board Advisor and author at The Scale Readiness Ladder, Scale Readiness Lab (United Kingdom)
“The most expensive mistake I see is automating a job nobody had defined.”
“So "human or agent" is the wrong question. The right one is "have we built the system yet?”
He moves the debate from the tool to the system underneath it. Build that first, and the hire-or-deploy choice gets a lot less dramatic.
No process, no handoff
João Rios, advisor to CEOs and founders (Brazil)
“Delegation is not about volume, it is about sequence. What never became a process cannot be handed to anyone, person or software.”
“The expensive mistake is treating a priority problem as a method problem. A new tool without a criterion just gives every demand the same weight.”
A Brazilian advisor puts sequence ahead of speed. A shiny tool can't tell you what matters first.
Where they split
Mitchell would remove the step before he'd automate it. His proof is a clinic's booking numbers, not a theory.
Paul would define and measure the work first. Then, in his words, "the repeatable, rules-based part goes to the agent".
João goes a layer lower: decide what matters and in what order, then pick the tool.
Nobody here is against agents. All three put the agent second.
The differences are small: one says delete, one says define, one says prioritize. Each is a question to answer before anyone gets hired or deployed.
My take
AI is brilliant at doing a defined task fast. That's also the trap, because a broken task done fast is just a bigger mess. Before asking whether to hire or deploy, ask whether the job is written down well enough to hand to anyone at all. If it is, let the agent take the repeatable parts and keep people on the judgment, relationships and accountability.
Your turn
What step in your business turned out not to need a person or an agent, just a delete key?
Next: agents owning pipeline qualification helped a sales team close 40% more deals, but automating investor relationships lost two deals in 60 days, and two other founders keep the close and client relationship human.
