I was at a conference this week with a ton of Vistage Chairs, folks that interact with CEOs across the country every day. And I had more than one conversation with folks who are hearing the same thing from CEOs everywhere: “we're doing something with AI but not sure it's the right thing.”
That puts them ahead of the 80% still stuck in “we should look into AI” mode.
But here's what I keep seeing: the first instinct is to survey staff about what AI tools they're using and what tasks take time.
That tells you almost nothing.
People don't know what they don't know. Someone might say “I use ChatGPT for editing emails,” but that doesn't surface the six-hour monthly report that's begging to be transformed. Asking about tools gets you a list of tools. It doesn't get you leverage.
Workflows, Not Tools
Instead of starting with tools, start with workflows. Specifically, look for anywhere your team is acting as a translator, taking information from one format or system and manually converting it to another. That's where AI creates leverage.
Your accounting team says reconciliation takes time. But the question isn't “can AI help with reconciliation?” The question is: what exactly are we reconciling, between what systems, and what decisions does that information feed?
Once you understand the workflow, the tool choice becomes obvious. Start with tools, and you'll spend six months experimenting. Start with workflows, and you'll know exactly what to build.
The Better Question
Here's what I'd suggest for your next planning session: instead of asking “what tasks take time,” have teams identify “what decisions do we make repeatedly, and what information do those decisions require?”
That shift changes everything.
One leader I shared this with extended it further: we should also be identifying the decisions we don't make, the ones we default, delay, or skip because we don't have time to do the analysis required to be systematic and thoughtful.
That's the real unlock. AI doesn't just accelerate the decisions you're already making. It makes visible the decisions you've been avoiding because the information cost was too high.
What This Looks Like in Practice
A nonprofit leader I know was preparing a concept note for a foundation. She could have asked ChatGPT to “help write a grant proposal”, the generic prompt that produces generic output. Instead, she went back and forth with the tool to align her ideas, in her words, to the foundation's priorities.
The AI did the deeper dive on priorities for her. She could cross-check its analysis against the actual call for proposals. She got the concept note done faster and better than if she'd done multiple rounds herself or with her team.
Notice what happened: she didn't use AI to write faster. She used it to think more thoroughly. The tool synthesized information she hadn't fully processed herself. That's the difference between AI-as-typist and AI-as-thinking-partner.
The Question to Bring to Your Next Meeting
Walk in with this: What decisions are we making, on repeat, that require information we're currently synthesizing by hand?
And the follow-up: What decisions are we not making at all because we don't have time to gather and analyze what we'd need?
The answers to those two questions will tell you more about where AI creates value than any survey ever will.