If the token leaderboards are full and usage is up and to the right, then everyone exhales.
The problem is those numbers answer the wrong question. They tell you how much AI is being used, but is it actually growing your business in the right way? Are your people getting better at the parts of their jobs that matter for making a real impact? I believe AI is most powerful working in concert with the people in your organization because people are the generative core.
So the question isn’t “how much AI are we using,” it’s “are we putting our people where they can do their highest-value work, leveraging the judgment and creativity that actually move the business?” When AI absorbs coordination, manual work, and first drafts, does that surface show up as more time on the work only people can do? Or does it just get eaten by more tooling?
What does this look like in practice? A GTM leader I’ve worked with measures AI not by usage, but by only one key metric: the share of her team’s time spent in front of customers. She found them at roughly 15% time with customers, with the other 85% going to coordination, admin, systems, and tracking. Her year-end goal is 50%+ of employee time with customers, and the team is going all in with AI with that goal in mind. It’s the cleanest example I’ve seen of measuring real business impact instead of AI vanity metrics. She’ll know AI is working when it frees people in her organization to exercise judgment, build stronger customer relationships, and spend more time where they create the most value. For her, that’s the clearest signal that AI is driving the business forward.
Following a familiar story in HR
The last time HR was promised liberation, it came shrink-wrapped as software. A better ATS, a smarter HRIS, an integrated LMS. The pitch was always the same: automate the busywork, free your people for the work that matters!
What actually happened is the tools multiplied, the integrations broke, and the data ended up scattered across a dozen systems that didn’t talk to each other. Some teams got to a genuinely better place with innovative tooling and solid People Operations after a lot of work. Many are still on that journey because SaaS systems didn’t make it easy to talk to each other because each wanted to be a big, all-in-one system. For those companies who couldn’t focus internal and external resources to optimize it all, the promise for HR never materialized, and the reclaimed time went to troubleshooting, maintaining the tools, and then teaching everyone else how to use them.
For most HR leaders, the promise of freedom quietly became machine-tending. The systems that were meant to help us instead got in the way of the real work. We spent less time actually talking to the people who needed us, and less time on the planning that keeps a team together.
And the cost was never only time. When the tooling makes HR slow, HR gets seen as the function that always says no, that can’t move at the pace of business, the cost center rather than the insight center. Sometimes the cost was concrete, like not being able to act on a new compliance requirement until we’d re-orged the data over days and weeks and then pulled the report to do it right. Sometimes it was the slow erosion of the function’s credibility. Either way, the systems built to help us were quietly making us look like the problem. AI now runs the risk of repeating the same pattern. The technology can work exactly as intended while still pushing people toward work that doesn’t create much value.
The trap is measuring the tool instead of the work
If the time AI frees up gets refilled with more AI, more prompt engineering, more monitoring, more dashboards to review, nothing has actually changed. You’ve swapped one flavor of overhead for another with a better interface. And if you start optimizing the whole organization around AI usage, you slowly hollow out the thing that grows businesses: people doing creative, connective work with other people.
Nobody plans this. It happens because hours spent in a tool are easy to count, and we aren’t yet in the habit of measuring the time humans get to spend on uniquely human work. So the easy number wins by default, and the systems expand to fill whatever space you let them. The risk is that we end up optimizing a more efficient version of the wrong thing.
Let’s do it differently this time
Let’s optimize for the work that grows a business: conversations with customers, planning, creativity, coaching, and good judgment. AI should create more room for those things, not become another system demanding attention.
This is the difference I saw in how we’re building Sol. Take something as ordinary as moving a person from one team to another. Today that small change usually can’t be finished by the manager who wants to make it. The tooling demands a specialist, a ticket, a hand-off, and context leaks out at every step. In Sol it starts from what you already know, that this person is moving to that team, and the system carries the thinking underneath it: the right effective date, the comp change that follows, who needs to hear about it and when. The manager doesn’t fill out a form. They get a system that already knows what they should be thinking about next.
You feel the difference in speed almost immediately. Insights that used to take months of manual work, re-orging data across systems just to answer a single question, now arrive in seconds. You can sit in a meeting with the answer already in your hands, at the moment you need it, instead of promising to circle back next week. This is what having a tool work for us looks like. This is the promise of the past delivered today, when work is changing and we need more time to leverage our human skills the most.
What to measure instead
Point the measurement at the human, not the machine, and build regular touchpoints so you can see change over time. A few questions that do that as you leverage more AI in your organization:
- For customer-facing teams: Is time with customers going up? Are relationships getting deeper?
- For builders and knowledge workers: Is time on creative work going up? Is AI easing tool management, or just reshuffling it?
- For HR: Is managers’ time with employees going up? Are HR leaders getting more room to plan, coach, and solve problems before they become issues?
- For leaders: Is your best people’s impact going up? Are they doing work now they couldn’t do six months ago?
These numbers sit a layer deeper than the surface measurements getting headlines around token maxxing. But they’re the ones that matter. The easy metric is what got us into machine creep the last time. Let’s not do that again. Measure the human impact on business outcomes, and we get ourselves back on track.
The part worth repeating
People are the generative force in an organization. That’s not a sentiment, it’s what I’ve seen over years leading the People function in business. The creativity, judgment, customer empathy, and instinct for what your market needs next; none of that lives in an AI model, and none of it shows up on an AI usage dashboard.
AI should give people more room for human work. If it pulls them deeper into machine tending, we haven’t changed the nature of work. We’ve just repainted the same walls.
We get to choose what this next phase of work becomes. The measurement is where the choice actually gets made because it drives decision making. It’s worth getting right. Building and adjusting your organizational design to drive outcomes with AI gets a lot easier when you have the right data to inform you and your leaders have the time to make the best decisions.
This is how the best companies will pull ahead. Not by counting tokens, but by multiplying human impact.
