If you’re leading AI at an enterprise level, you’ve probably heard this question more than once: “What’s the ROI?”
It’s the wrong question.
At the KPMG-organized GenAI Executive Accelerator in Boston last week, one thing became clear. Most enterprise AI programs are being evaluated with the wrong lens. Executives are chasing short-term validation when they should be designing for long-term advantage.
As Kevin Bolen, Advisory Principal and AI Transformation Leader at KPMG, put it: “What’s the ROI of email?”
Some technologies, like email, mobile, cloud, or GenAI, may not initially present obvious justifications with a clean business case. Rather, they redefine how business gets done over time. But only if the organization is willing to rethink how it builds, enables, and measures value.
‘Start with Personas’
Peter Salvitti, Chief Technologist at Boston College, said something that deserves to be repeated in every executive off-site: “Don’t build for use cases. Start with personas.”
That shift matters. Too many organizations are treating AI like a checklist of automation opportunities. But real value comes from building tools around the people doing the work, how they think, how they search, and what gets in their way. That also means you’re designing AI tools that are more likely to actually be used.
Chasing ROI or Driven by FOMO?
A recurring theme at the event: enterprises are stuck in pilot purgatory, too cautious to scale, but also afraid to fall behind. One attendee nailed it when he asked: “Are we playing to win, or just reacting to FOMO?”
And Ben Lavallee, Director at Google Cloud, added critical clarity: “Even for me, it’s hard to keep up with all the tech. Don’t obsess over the tools; focus on your people.”
Organizations’ AI agenda shouldn’t be centered on features or hype cycles, several speakers said. Instead, it should be centered on people, process, and trust.
Your Most Valuable AI Asset May Already Work for You
Some of the most actionable insights came not from vendors, but from operators.
Adam Landman, Chief Information Officer at Mass General Brigham, said, “It’s not just about hiring data scientists anymore. It’s about creating room for the builders already inside your org.”
He was talking about a pathologist at the hospital system who developed his own internal LLM.
And Satyendra Thakur, CISO at Hasbro, made it even simpler: “We try to keep it simple. Can GenAI help us grow? Can it help us upskill our people?” he said.
How States are Seeking AI Advantage
While many enterprises are waiting for federal guidance or debating pilot programs, Massachusetts is making a big bet on infrastructure.
Jason Snyder, Chief Technology Officer for the Commonwealth of Massachusetts, shared that the state is investing $31 million to upgrade a hydro-powered data center in Holyoke, Mass. It’s part of a broader strategy that includes:
- State-funded AI sandboxes for public agencies
- Collaboration with universities like UMass Amherst and Worcester Polytechnic Institute
- A public-private “data commons” to blend datasets across sectors
- Training standards for AI literacy and responsible use across government teams.
And it’s not just symbolic. One team of state employees, working in their spare time, built a natural language AI tool to help parents find childcare based on individual needs and preferences. It’s already saving weeks of manual work across multiple agencies.
“We’re not just funding ideas,” Snyder said. “We’re building capacity.”
What to Do Next
If you’re leading innovation, digital, or transformation, here’s where to focus:
- Drop the use case obsession. Start with personas.
- Shift from focusing on immediate ROI to focusing on friction. What’s slowing your people down?
- Build internal trust before you deploy external-facing pilots.
- Decentralize experimentation. Your best LLM projects might come from teams and departments you wouldn’t expect.
- Treat AI like crucial infrastructure, not a feature.
The Bottom Line
AI isn’t just another wave of tech; it’s the foundation for how future work will happen. And the companies that succeed won’t be the ones that roll out the most tools the fastest.
They’ll be the ones that reduce friction, redesign workflows, and train people to move ahead with confidence, not fear. In the end, the real return on AI won’t be measured by the number pilots launched, but rather how well your people can learn, adapt, and lead.
Editor’s note: KPMG has been a sponsor of InnoLead research and events, but this conference report is part of our ongoing editorial coverage of AI-focused events around the world.
You must be logged in to post a comment.