November 16, 2026 • 7 Min Read
At a building services company that turns architects’ drawings into plans that homebuilders then utilize, AI cut drafting tasks from four or five hours to four or five minutes.
Prior to the roll-out of AI, the company had been running behind on deliveries to builders. Now it was getting plans to them on time far more often.
When the AI roll out began, the company found that junior designers were among the slowest to adopt the tools, because they could see that the work they had been hired to do was evaporating. That’s when the company told employees they wouldn’t lose their jobs because of the rollout. Instead, it would use the hours saved on drafting to prepare them for more advanced assignments.
We’re replacing all those hours with training, and eventually we think we will get more value in the long-term.
- Building services company HR leader
At an RGP roundtable in Houston, finance, technology and HR leaders gathered to discuss what businesses were getting from their investments in AI. In RGP’s recent research on AI x Humans, 73% of respondents said they get more done in a typical week than they did before using AI. Eighty percent said the time they saved was simply filled with more work. That raised a question for the leaders gathered. When people get more done, what improves for the business?
Some executives said gains are visible early, as when a team starts meeting more delivery deadlines. Others take longer to judge, especially when a company uses the time it saves to train people for work they couldn’t do before. The return depends in part on how the company changes the way it works and what those changes deliver over time.
AI can cut hours from a task. The return depends on what teams do with the time they get back and how they make sure the work is right.
What Can Be Measured Today and What Can’t be
It became clear in the discussion that some ROI can be measured quickly, and some ROI remains elusive.
At a small public company, the finance team received roughly 800 to 1,000 freight invoices a month. Before paying one, someone had to confirm that the shipment had taken place and that the charge matched the contract. An employee volunteered to build an AI workflow that could read the invoice, verify the shipment and compare the charge with the company’s contract data.
The workflow saved about 80 to 85 hours a month. As people left, the finance team reassigned their work and kept up with the same volume without filling the vacancies. That gave the executive an ROI result to take to the board. The company could handle the same workload with fewer people.
The board wanted to know how the team would catch mistakes. The tool checked invoices before payment, and the company’s existing system still handled the transactions. Finance added a manual reconciliation to check that the tool’s results could be relied on. The executive estimated that the additional checks took about two hours. “We’re still up,” the executive said.
Another participant raised a cost issue that could be harder to estimate upfront. An AI workflow had gone down after employees had left, and the company had to find people to take over and clear the backlog. For a finance team considering a similar change, the amount of work it can handle with fewer people is one figure it can measure today. The cost of restoring manual coverage may only become clear when it needs it.
What happens as more employees begin using AI? A finance leader at a software company described how the company wanted people to find new uses for the tools, and then it had to reckon with what that use was costing. “We want everybody to do it, we want it deployed across the enterprise, and then the bill comes in and you’re like, ‘Let’s turn it down a little bit.’”
A supply chain executive offered a fascinating perspective. His team manages thousands of products. AI can now quickly flag where margins were slipping or where inventory was piling up. He has re-designed work for his team so they can spend its time finding out why. He believes that’s the work that might lead to a key decision about pricing, purchasing or which products to carry.
“I’ve learned that productivity is important, but don’t fill up the white space. The white space is important too and where greater value can ultimately be unlocked,” he said.
The People Side of AI Use
A CFO described how less experienced employees could use AI to put together polished presentations quickly, but the CFO still found that he had to scrutinize the figures behind the slides. The company now holds weekly demonstrations where employees show colleagues how they use AI. Those sessions help people learn what the tools could do and how to tell when an answer is wrong.
Jones described how RGP’s finance team gives employees practice catching errors. The team feeds AI an outdated report or other flawed data and asks employees to work out what went wrong. The employees learn to trace the numbers back to their source before someone makes a decision based on them.
One partnership in which greater speed was achieved prompted a closer look. A finance leader at a software company described how his company had been considering a partnership with another technology company to serve a large energy client. When his technology team and the partnership’s technology team compared how they worked, the prospective partner’s team was making six times as many changes to its codebase per person.
“How can you possibly be reviewing the quality of that code?” the executive recalled his head of technology asking.
The software company walked away from the collaboration after the prospective partner failed to show it had enough people to review all of its code. The executive’s company is now hiring developers. And more of their time goes toward reading code and deciding whether it does what the business needs. At the end of the day, the company has gotten faster at writing software, but it still needs people who can make sure that software is ready for a client.
In RGP’s AI × Human research, 63% of respondents said they often or always disclosed their own AI use, while 47% believed their colleagues did so as often. For someone signing off on a client deliverable, that gap raises a practical question. How do they know where AI was used and what has already been checked?
I’ve learned that productivity is important, but don’t fill up the white space. The white space is important too and where greater value can ultimately be unlocked.
- Roundtable supply chain executive
As the discussion wrapped up it was clear that there is still a lot of complexity in trying to measure ROI. Yes, companies can see the first benefits of AI in productivity. More orders can go out on time. A team may be able to handle the same workload with fewer people. But the other, potentially game changing gains need a longer time to develop. If AI takes over the assignments that once taught people the basics, it may take months to see whether a new approach to training prepares them for more demanding work. Leaders should decide what to improve first and what the business will have to change for larger and more lasting ROI. And the answers may change once employees start using the tools.