Sydney, Chicago and Cape Town on the trust cost of over-automation
A Cape Town media director, a Chicago legal-marketing co-founder and a Sydney consulting founder agree AI can take the prep work, and warn about what speed quietly costs.
Faster replies. Worse conversions.
That is the trade one business made when it automated first contact, according to a media director in Cape Town.
Two leaders in Chicago and Sydney land in the same place from different directions: the expensive part of automation is not the tool. It is what you stop paying attention to.
We asked leaders which parts of a job AI has actually taken, and which still need a person. These three answers share one worry: speed that costs you the customer.
Admin went to the bots. The buyer's signal got lost.
Dieter Blom, Director of Versys Media (Cape Town, South Africa)
“We’ve cut admin time by more than half with automated scoring and routing, but sales teams still make the final call on deal quality.”
“One business reduced response time sharply but saw conversion quality drop because important buyer signals were being flattened into generic replies.”
Scoring and routing are routine, so they moved. In one service business, workflows ran first-touch replies, enrichment, booking logic and CRM updates end to end, and admin time fell by roughly 60%. Reading what a buyer actually meant did not move well. The speed metric looked great while quality slipped.
The AI drafts. A human signs.
Sasha Berson, Co-Founder and Chief Growth Executive of Grow Law (Chicago, United States)
“AI performs all tasks of data collection, analysis, and generation of the first draft at a fast pace. However, the final verdict is still up to human beings, who need to make sure that they verify the content and confirm that the produced content represents the correct version.”
“The lost benefit of over-automating processes is not missed accuracy but a lack of trust and reputation which are more expensive to gain back.”
Collection, analysis and first drafts are the labor, and the labor is cheap now. The pricey mistake is skipping the human check, because trust is slow to rebuild.
Prep went to AI. The decision stayed put.
Joe Papadatos, Founder, Strategic Growth Advisor of Two Icons Consulting (Sydney, Australia)
“just because AI can generate an answer doesn’t mean it should become the answer.”
“The better question is: what should AI enable this organisation and its people to do better?”
Research, summaries and drafts are covered. What still needs a person is ownership: someone has to answer for the result. His answer to the human-or-agent debate is to change the question, and the new one is about people, not tools.
Where they split
Nobody here argues for fewer machines. All three keep a person on the final call.
The real disagreement is small. It is about where the damage shows up first, not whether it is real.
My take
Automate the work that prepares a decision. Keep a person on the work that carries a consequence. Before you automate a step, ask what signal it throws away. If you can't name one, you haven't looked hard enough. Speed is easy to measure. Trust only shows up later, and usually on a bad day. So put a human at the one point where a wrong answer would cost you a customer, and let the machines have the rest.
Your turn
Which step in your own process got faster and quietly got worse, and how did you notice?
Next: an SEO founder, a content creator and an automation engineer all approve AI output before publication, but the engineer adds a different worry: the knowledge you expose to the provider.
