A friend who owns a PR firm sent me a link in December and said, "Check out this really cool tool."
I set up an account without really knowing what it was. My first prompt was: "Tell me about SSO and how it works."
That might seem like a strange place to start. But at the time we were rolling out a new product feature, and the handoff from development to marketing had been sparse. No real use cases. No messaging. No positioning. Nothing that helped explain why the feature would matter to customers or how it might make their lives easier. I figured I'd ask ChatGPT and see what came back.
It responded within seconds. My immediate reaction was: that's genuinely useful.
Not because I suddenly thought AI was going to replace everything. But because I could immediately see how it fit into the kind of work I was already doing.
What I Was Actually Using It For
My approach was practical from the start. I wasn't treating it as an all-knowing machine or asking it to do the work for me. I was using it the way you'd use a very fast, very patient thinking partner.
When I needed to understand something quickly (especially technical information that needed to be translated into something clearer) it helped. When the information I'd been handed was incomplete or framed in a way that made it hard to work with, it gave me a starting point. When I needed to organize my thinking, explore an idea, or pressure-test how something might be understood from a customer's perspective, it was useful there too.
In the SSO case, I wasn't looking for something I could paste into a launch email. I needed a faster way to get my head around the topic so I could start asking better questions. What does this actually solve? Why would a customer care? What kind of friction does it reduce? How should it be explained in a way that feels clear rather than technical?
That's a much better use of the tool than asking it to produce polished content and hoping for the best.
Where the Real Value Was
What I kept noticing was how useful it was in the early messy stages of work. The moments before the work is shaped: when you're trying to understand, clarify, and connect the dots. That's where the friction usually is, and that's where I was finding the most value.
It also made me think differently about knowledge work in general.
In a typical workday, the real bottleneck is often not skill. It's time and friction. The time it takes to dig for information, translate complexity, or get unstuck when something hasn't been fully thought through by the people handing it to you. ChatGPT reduced that friction in ways that felt immediately practical.
That doesn't mean it's always right. It doesn't mean it replaces expertise, context, or judgment. What it means is that it can move the work forward faster in the moments where speed is what's actually missing.
That's what caught my attention in January 2023. Not the grand claims. Not the predictions about what AI would eventually be capable of. Just the practical observation that this thing could help me make sense of things faster, right now, in work that was already in front of me.
Everything since has been an extension of that.