AI transformation succeeds or fails on one decision most leaders never make deliberately: what happens to your people once you take the busy work off their plate?

Dustin Williams built his strategic outcome engineering approach inside Google’s data and product team, where he helped structure massive partnership deals by identifying where a business was actually losing value. Now, Dustin brings that thinking into private equity, family offices, and search funds. His argument is simple: automation doesn’t fail because the model is wrong, it fails because nobody thought about what happens to the person the model replaced.

“That is you putting them up on stage, telling them they’re a rock star, and giving them a chance to shine.”

Businesses that succeed with AI transformation projects do so because they make their employees feel valued. Those businesses achieve this by repositioning employees to more prestigious positions where they can add more impact and value. The businesses that don’t, the employees feel threatened and slow, if not outright stop, the AI transformation.

Dustin’s fix isn’t sentimental, it’s mechanical. Find the task that costs the business the most in unproductive or wasted hours, use AI to automate it and redirect the employees’ recovered time toward work that actually grows revenue. This has several outcomes: cost savings, increased revenue / productivity and an engaged employee who wants to see the project succeed.

When an AI automation of this nature is priced honestly, that kind of friction is rarely small. A habit as small as arguing over who should send an email can cost a business a million dollars a year in person hours alone.

The saving was never the point. That same million dollars, spent instead on freeing those people to call the best customers and ask for the business, made thirty million dollars a year. The people weren’t replaced, instead their job was upgraded from busy work nobody wanted to the kind of work that actually moves revenue. That is the whole mechanism: take away what they hate, hand them something that makes them feel like they are moving the business, and the resistance that kills most transformations simply never forms.

The starting move isn’t complicated. Ask your people directly what they never want to do again, not their KPIs but the actual task they complain about after work. That is where you start looking, because the work people resent is usually the work that is quietly costing you the most.

Melissa’s Take

I’ve watched this go wrong more than once, and I’ve also lived it from the other side of the announcement. I was part of a Copilot deployment where I found out I had a license after the fact, nobody told me, I just discovered it was sitting there. The company never asked a more basic question first: who’s actually using these, not who should be using them on paper, who actually is.

That’s the exact gap Dustin is describing, just in miniature. The tool showed up, the announcement probably happened somewhere, and the part that never happened was anyone connecting the rollout to an actual person’s actual workday. No one asked what I’d want automated. No one asked what I was doing with my time that a license like that could have freed up. It just existed, quietly, and whether it got used or not was left entirely up to whether I happened to notice it and figure out the value myself.

That’s the difference between deploying a tool and leading a transformation. A deployment is a license getting assigned. A transformation is someone asking “what’s eating your week” before the tool ever shows up, so the rollout lands on a real problem instead of landing on nothing. Dustin’s version of this costs a company a million dollars a year in wasted arguing over who sends an email. Mine cost a lot less dramatically, just a license nobody explained and nobody measured, which is its own kind of expensive. Unused capability isn’t neutral. It’s a signal to your people that the investment wasn’t really about them.

Here’s the gateway action, and I’d add a second one to Dustin’s: this week, ask your people what they’d never want to do again, and write down the answer. Don’t fix it yet, just find out what it is. Then go check who in your company already has access to a tool like this and isn’t using it. That gap will tell you more about your rollout than any adoption dashboard will, because it shows you exactly where the “why” never got explained.

This is part of a full conversation I had with Dustin on the Executive Connect Podcast, which goes deeper on how he uses AI to catch deal breaking red flags before due diligence even starts, why he thinks most AI agents are given far more autonomy than any business actually needs, and the one mindset shift he believes every PE leader has to make this year.

Each month, I publish The Full Brief which examines these conversations in more depth, provide my own insights plus take-away points you can action immediately. When you join The Full Brief, you will also unlock the Maverick Test, a three minute quiz that reveals your Maverick Archetype and what it means for the way you lead, decide, and build.