.png)
Is your organization's AI investment actually driving results, or just adding another line item to the budget?
That question shaped the conversation at AI in Action: How Leading Companies Are Actually Using It to Drive Results, a panel discussion held August 20, 2026, in Omaha, Nebraska, and hosted by the NACD Heartland Chapter. The National Association of Corporate Directors' (NACD) Heartland Chapter brings together board members and executives across Nebraska, Kansas, Iowa, Missouri, and Oklahoma to discuss the challenges shaping today's boardrooms.
The panel featured:
Sandra Hulm, Senior Vice President at Scoular, moderated the discussion.
(From left to right: Sandra Hulm, David Tomlinson, Justin Webster, Danielle Egr, & Brody Deren)
The panel covered what's actually working with AI, what isn't, how to measure its value, and what boards should be asking their leadership teams. Below, we break down the key takeaways and answer some of the toughest questions from the room.
The panel covered seven themes: real business outcomes, scaling past pilots, measuring AI's value, budgeting for AI costs, platform strategy, workforce skills, and the board's governance role.
A few questions sparked the most discussion in the room: how companies actually scale AI past the pilot stage, how they prove it's worth the investment, and why AI is so hard to budget for. We break all three down below.
The biggest difference comes down to two things: data readiness and workflow integration.
Companies stuck in pilot mode tend to treat AI like a software license. They buy some seats, hand them out, and hope people figure out how to use them. Companies that scale successfully treat AI as a bigger business shift. They build it into how the work actually gets done.
Tomlinson saw this play out firsthand. When his team rolled out an AI tool company-wide, usage was lopsided. Just 10% of employees accounted for half of all activity. Giving people access wasn't the same as getting them to change how they worked. "Creating awareness does not create desire," he said.
Data readiness matters just as much. Webster and Egr both described AI projects that struggled for the same reason: the technology worked fine, but the data underneath it wasn't reliable. Egr's team built a tool to understand why customers were calling in to their call center, only to discover it was misreading the real problem entirely, because the underlying data wasn't accurate. Webster had a similar experience building a decision-support tool on data that turned out to be messier than anyone realized.
Deren pointed to a similar pattern with a homebuilder client. Its CIO had rolled out AI licenses to hundreds of employees, but when the board asked what results it had produced, he didn't have a clear answer. What made the difference was the company's earlier investment in a modern, well-organized data platform, which gave Trility something solid to build a real workflow-based AI solution upon. From there, the work moved from a generic tool to something built around a real business decision.
"If you are distracted by shiny new AI models and ignoring your proprietary data plumbing, you will stay stuck in pilot mode," Deren said.
Track outcomes, not activity.
Many companies fall into the trap of measuring the wrong things, like how many people are using an AI tool or how many lines of code it wrote. Those numbers feel productive, but they don't tell you if the business is actually better off.
The better approach is to connect AI usage to metrics that already matter to the business: speed, quality, and revenue. In software engineering, Trility points clients toward DORA metrics, an industry-standard framework that measures how fast code moves from request to production, how often releases break, and how quickly bugs get fixed. If those numbers aren't improving, the AI tool isn't earning its keep, no matter how many people are using it.
The same logic applies outside of software. A claims team might track how many claims get processed per month. A customer service team might track resolution time. The metric changes, but the principle stays the same: if AI usage doesn't lead to faster delivery, higher quality, or more revenue per employee, it's just noise.
Because the sticker price you approve is only a fraction of what you'll actually spend.
Traditional software has a predictable cost: a per-seat license fee. AI doesn't work that way. Usage-based pricing means the bill can swing wildly depending on how much people actually use the tool.
Panelists are already adjusting how they track it. Tomlinson pushes AI costs back onto the business units using them, so the people spending the money are the ones deciding if it's worth it. Webster expanded his team's old asset-tracking function into what he calls "Technology FinOps," giving business stakeholders real-time visibility into spend. Egr described a similar shift at Nelnet, from leadership absorbing AI costs centrally to splitting them back out to business units as usage matures. "I feel like to this point we're all toddlers touching a bunch of stuff and just figuring out how it's working," she said. "It's time for us to mature now."
Deren broke the problem into three cost channels, and noted most boards only see the first one: the platform and model fee itself. The second is hidden inside software companies already own, as vendors bundle AI features into existing plans and raise prices accordingly. The third is what employees are already expensing on their own; Deren pointed out that ChatGPT has become the most-expensed application inside the average enterprise, ahead of both AWS and Microsoft 365.
"Your AI spend isn't your AI vendors' invoices," Deren said. "It's those, plus an AI tax spread across a few hundred SaaS contracts that nobody is adding up, plus whatever your people are already expensing."
Across every question the panel tackled, one theme kept surfacing: the companies getting real results from AI aren't the ones with the flashiest tools. They're the ones with clean data, clear workflows, and a way to measure whether any of it is actually working.
That's a harder problem than buying a license. But it's also a more solvable one, and it's the same discipline that separates a pilot from a company-wide win.
Every panelist agreed: the companies winning with AI aren't guessing. They're measuring.
If your leadership team is asking for proof that your AI investment is paying off, and you don't have a clear, data-backed answer yet, you're not alone. It's the exact problem one Trility client faced before using a DORA-aligned diagnostic to turn vague pressure into a clear, prioritized plan. See how that played out.
Want to talk through where your organization stands? Get in touch with our team.
