Not in a survey.
In conversations — 60+ of them, recorded, transcribed, and eventually turned into my book - The Science of GIS Leadership.
The question was simple: what are you most proud of?
I expected to hear about projects. Programs built from scratch. Budgets fought for and won. Systems that outlasted the politics around them.
That is not what I heard.
Every leader I interviewed — without exception — pointed to a person.
Someone they had developed. Someone they had believed in before that person believed in themselves. Someone who went on to lead something the original leader could never have led alone.
I have been thinking about this pattern a lot lately.
Because right now, the conversation in this community — and in every GIS organization I am in contact with — is about AI. About what it can do. About what it is replacing. About what happens to GIS professionals when the technical tasks that defined their careers become prompts.
It is a real question. It deserves a real answer.
And the answer I keep coming back to — the one I heard in 60+ interviews before any of us were talking about agentic AI — is that the great GIS leaders were never defined by the tasks they could execute.
They were defined by what they built around them.
I had a conversation this week with Gary Maguire — the inaugural Australian GIS Hall of Fame inductee — that confirmed it again. He said something I want to put in front of this community:
"As leaders, it is our responsibility to leave a legacy — not defined by the GIS platform or solution we built, but by the people we developed and lifted as the next generation of GIS leaders."
Gary said that at the end of a 90-minute conversation that covered integrity, trust, generosity, and what it actually means to lead well.
AI was not mentioned once.
And yet that conversation answered the AI question more clearly than anything I have read in an industry report this year.
So before July 12th — I want to ask this community directly:
What does a great GIS leader do that AI cannot?
Not what they should do. What do the great ones actually do — that no prompt, no model, and no agentic workflow could ever substitute for?
I would genuinely love to hear your answers before we meet in San Diego.