Easy to automate, hard to automate well

Hi {{first_name|there}},

Most companies have bought the AI seats by now. Sawyer Middeleer works on the harder question that comes next: has anyone actually started working differently? His answer starts with the Chief of Staff, and with being honest about which work shouldn't be automated at all.

Also this week: Chief of Staff Connect New York is eight days out, and our Australia chapters are closing out their first year with end-of-year meetups in Brisbane, Melbourne and Sydney.

Sawyer Middeleer was the first hire and Chief of Staff at Aomni, where he ended up shipping production code from customer calls. Now he helps companies that have bought the AI seats figure out why nothing has changed.

Picture the support ticket. A bot can close it for five cents. A person costs five dollars. The math looks obvious. A lot of AI mandates are built on these exact assumptions.

In Sawyer's experience, it's rarely that simple. The person closing the ticket was never just moving text from one box to another. They were carrying context nobody had written down.

So his starting point isn't a tool. It's the process. Map how the work gets done today, where the value comes from and what a person carries into it that nobody wrote down. Then ask whether that context can go into the AI version too. Where the AI's output and the person's output look the same, hand it over. Where the person has, in Sawyer's words, "that special something," keep them in.

That's why he likes to start with the Chief of Staff. Most companies have bought the seats, and adoption metrics stopped being interesting once everyone had the tools. The open question is whether anyone works differently, and the Chief of Staff usually has the clearest view of the processes, the stakeholders and where the resistance sits.

It also reframes the mandate most Chiefs of Staff are getting right now: grow without hiring. Sawyer takes revenue per head seriously as the clearest payoff of AI done well. He's just as clear about the trap.

Brute-force the time-saved number across everything and you get slop: more bugs, weaker marketing, a company slowly losing touch with its customers.

The harder part is that the line between what AI should and shouldn't touch keeps moving. Something that was a bad idea to automate six months ago may be fine today. Which raises the question every Chief of Staff running this work will face: who in your company is responsible for noticing when that line moves?

Sawyer leads the "Building the AI Flywheel" workshop at Chief of Staff Connect New York, October 22 to 23. Limited tickets available!

October 22 to 23 at 151 West 42nd Street, with 150+ senior Chiefs of Staff. This year's theme is From Force Multiplier to Builder: the shift from amplifying leaders behind the scenes to owning the systems, programs and operating models that move the business.

Sawyer's Building the AI Flywheel workshop is one of the sessions, and the full agenda and speaker lineup are on the event page. If you've been weighing it, this is the week to decide.

October 22 to 23 · 151 West 42nd Street, New York

Sawyer and I ended on the part of AI work that never shows up in an adoption dashboard: deciding where people stay in the loop, and revisiting that call as the tools get better.

If that's the conversation you're having inside your company right now, bring it to New York next week. His workshop is built for it.

Scott