The Interesting Part Isn't Always the Automation
Collin Michels3 min read

Each week I get to see how brokers and also the team at OneSource are using AI day to day. It’s usually less of the huge flashy systems being built (although this does happen) and more of unique ways they can work out a problem that most of the time either is really complicated or takes too long to do on their own. It reminds me that although automation is a large part of practical AI use, it’s not always the most interesting part.
Marc Zao-Sanders has been tracking real AI usage patterns for a few years now through his “AI in the Wild” research, and he breaks down what he found in Thinkslop, backing up what I hear from the people around me. The way lots of people use AI looks different from the way it is often pitched.
Here are my takeaways:
- Troubleshooting is one of the fastest-growing use cases for AI. Not just “write this for me,” but “help me work through this specific, stuck problem,” in real time.
- People are handing over more of the actual work, not just the first draft. Multi-step, tool-using tasks barely showed up in last year’s data. Now they’re near the top.
- The value can be in the back-and-forth, not always the finished product. A lot of this usage looks less like delegation and more like thinking out loud with someone who talks back.
None of that surprises me when I look around. The brokers getting the most out of AI aren’t using it to skip the prep; they’re using it to do more of it — reasoning an objection, better client research, presentation repetitions. It’s not replacing the work of Mastering Your Tone. It’s a rehearsal room for it.
Charlie Coppola
Connect with me on LinkedIn!
Thinkslop
intention, agency, self-possession, cognitive sovereignty
Jun 03, 2026
My career is an odd one. My first job was strategy. Then start-ups, a charity, a social enterprise, founded filtered.com, wrote Timeboxing and, this year, started AI in the Wild. I’d not reflected much on all these decisions and exertions until writing this newsletter. There are some common threads running through these various endeavours: intellectual challenge, societal benefit, learning & education, doing something that’s not been done before, and autonomy.
And intention. Intention is at the intersection of timeboxing and AI. It’s essential to timeboxing because to timebox anything, you need to choose what matters most to you, what you want and need to do, what you intend. It’s essential to AI because as more and more of us have this powerful tool, we need to tell it what to do for us - we need it to fulfil our intentions, and for that we need to be crystal clear on what those are.
But our intentions have become opaque to us. There’s so much going on to confuse, overwhelm and influence us. Algorithmic feeds, advertising, tribalism, notifications, propaganda, the opinions of others, content overload, etc. It’s hard to know what we even really think these days.
On Monday of this week, Harvard Business Review published this article that I wrote. It’s a summary of the third annual study into how people are using AI. One of the main themes is that people are worried they are turning to AI too easily, giving up on the thinking process themselves. Note that thinkslop is not output (like what is meant by AI slop or workslop). Thinkslop is a cognitive behaviour, a cognitive laziness. And it’s bad because we deny ourselves the fun of thinking, we may well be denying ourselves the possibility of having certain creative thoughts, and over the long term, we may even be eroding our ability to think.
And AI is just one of the many modern phenomena that inhibit independent, critical thought and actions.
In the HBR I didn’t offer anti-thinkslop solutions. It’s difficult because the system we’re part of thrusts us more and more into unthinking behaviours. The first step must be to become aware of the problem and to want to reject it. The second is to develop a system to guard against it. Timeboxing is such a system - repeatedly and reliably giving yourself a little time and space to become aware of what’s most important to you. With AI, it might be to restrict the situations in which or the degree to which you use it. With algorithmic feeds, it might be to turn some of them off. With notifications, it might be to permit fewer or permit them only at certain times.
Find a little time and space to reacquaint yourself with your intentions, reclaim your agency, and think for yourself.

