A few weeks ago, I was helping a team sort through the aftermath of a kickoff meeting. You know the kind: everyone shows up prepared, the conversation is productive, and by the end, there’s a shared sense that the project is moving forward.
Then the meeting ends, and reality sets in.
Half the team heard slightly different versions of the goals. A few requirements were mentioned casually but never confirmed. Someone referenced a timeline that sounded reasonable in the moment, but suddenly feels optimistic when you actually think through the work.
This is a normal moment in project management. It’s also the kind of moment where AI is suddenly very tempting. You’ve got a transcript, notes, and a pile of information that could easily turn into a project brief or plan if it were properly organized. So the instinct is simple: paste everything into an AI tool and ask it to generate the documentation.
The result usually looks impressive, and your AIbot will tell you it was a great idea. It'll generate your document with a structure that resembles a thoughtful project artifact, complete with bullets, subheads, and real data from your meeting.
But when you actually dig into what the document is saying, you start to notice something: it looks better than it really is.
On the surface, AI delivers a document that feels organized and complete. But when you compare it to what actually happened in the room — the side comments, the hesitation around certain ideas, the places where people were clearly interpreting things differently — the document doesn’t quite capture it.
That’s because our job isn’t just organizing information. It’s interpreting it.
A project manager listens for the context behind what people say. We notice when a requirement sounds settled but still needs clarification. We recognize when two stakeholders are describing slightly different versions of the same goal. A lot of the work happens in that space between the words — in the judgment calls we make as we translate conversations into direction for the team.
AI isn’t interpreting that work. It’s taking the input and organizing it into something that looks coherent — clean sections, neat bullets, a structure that feels convincing.
And once you start noticing that difference, the question of AI becomes much more interesting. Because suddenly the decision isn’t just how to use AI on a project. It’s deciding when it actually belongs in the process.
The simple filter I use before opening AI
Over the past few months, I’ve started using a simple mental filter to make that call. Before opening an AI tool, I ask myself one quick question: Am I trying to figure something out, or am I trying to organize something I already understand?
That small distinction changes how I approach the entire workflow.
Sometimes a project manager is doing thinking work. Other times the work is about structure.
AI happens to be excellent at one of those.
The difference between thinking work and structure work
Project management constantly moves between two modes.
