There’s no shortage of success stories about AI helping people work faster. Research, analysis, writing, and preparation that used to take hours now take less. But as those stories pile up, one question keeps coming back to me: when AI saves time, where does that time actually go?
Scattered savings aren’t capacity (yet)
Most AI time savings show up as 20 minutes here and there. Those minutes are valuable. They might help someone catch up, improve the quality of their work, or create a little breathing room. But 20 minutes saved across 100 people doesn’t automatically become capacity you can assign to another project.
Those savings are spread across different roles, schedules, and points in the workday. Before we can put that time to work, we need to understand where it exists, whether it’s repeatable, and what work it can realistically support.
When the savings become visible
Some savings are easier to see. For example, when a team consistently needs less preparation time for client workshops, even after reviewing and revising what AI produces, that changes the effort required to deliver the work. If those gains are repeatable, they should eventually influence how future engagements are estimated, scheduled, and staffed. The work has changed, but the plan likely still assumes the same effort and the same assignments.
That gap is common. In a new IBM Institute for Business Value study of more than 1,600 senior AI executives, only 29% said their organizations quickly adjust decisions and resources when AI performance data points to an opportunity. Another 40% said those opportunities get discussed, but action is delayed. In other words, organizations are identifying opportunities, but turning those opportunities into timely decisions remains a challenge.
Three questions to ask
- Is it repeatable? Will the savings show up project after project, or is it a one-time win?
- Is it showing up in the numbers? Are estimates and actuals reflecting the change, or does the plan still assume the old effort?
- Who owns the decision? Can the team reinvest the time on its own, or does the trade-off cross projects?
What closing the gap looks like
Some teams are already closing it. Jennifer Elkins, RMCP, shared an example of a team that used AI to speed up report writing, then revisited its delivery schedules to get deliverables into clients’ hands sooner.
Notice what they did with the time. They didn’t free someone up for another project. They delivered faster. That distinction matters, because AI productivity can create value in several ways:
- More capacity: taking on additional work or helping another project get back on track
- Faster delivery: getting results into clients’ hands sooner
- Higher quality: more time to test, refine, and review
- Stronger client outcomes: more time with the client
- Team development: more time to mentornewer team members and build capability
Not every hour saved needs to be reassigned. The real opportunity isn’t simply recovering hours. It’s making better decisions about how the organization uses the productivity it gains.
Where Resource Management fits
The people doing the work are often the first to recognize what AI has changed. Project and engagement managers can determine how those changes affect their plans. Practice leaders can update estimating standards and delivery methods so future engagements reflect what’s been learned. Resource Management brings a broader view, helping organizations recognize where productivity gains create new options across projects, priorities, and workforce capacity. It can then help leaders translate those options into deliberate decisions about how resources are planned and deployed.
That’s a real opportunity for our profession. As AI changes the work, the plan has to keep pace, and Resource Management is where the two come back together.Consider where AI has changed the work on your projects. Has the plan caught up, or does it still assume the old effort? Where do you have gains that are repeatable enough to plan around? Are you deciding how to use that productivity, or is it being absorbed by default?







