The quiet revolution happening in offices everywhere—and why it matters more than you think
By Michael DiNapoli
Something strange is happening in workplaces around the world, and most people haven’t noticed yet.
At Danfoss, a global manufacturer you’ve probably never heard of, customer emails that used to take 42 hours to process now get handled in near real-time. At Telus, employees are saving 40 minutes every time they interact with a new kind of colleague. And at Suzano, the world’s largest pulp manufacturer, questions that used to require a data analyst and half a day now take seconds.
The common thread? None of these efficiency gains came from hiring more people.
They came from AI agents—and 2026 is turning out to be the year they went from science fiction to office reality.
What Exactly Is an AI Agent?
Let me be clear about what we’re talking about, because the term “AI” gets thrown around so loosely it’s almost meaningless.
An AI agent isn’t ChatGPT. It’s not a chatbot that answers questions. It’s not autocomplete on steroids.
An AI agent is software that can actually do things. It can log into your company’s systems, pull data from multiple sources, make decisions based on rules you set, and complete multi-step tasks without someone hovering over it.
Think of it this way: a traditional AI tool is like a very smart reference librarian. You ask a question, it gives you an answer. An AI agent is more like a new employee who happens to work 24/7, never gets tired, and can access every system in your organization simultaneously.
Here’s a real example. A real estate company called JBGoodwin built an AI agent that pulls housing market data every week, writes an 800-word blog post analyzing the Austin market, and emails it to the team for review. Every week. Automatically. The humans just approve and publish.
That’s not answering questions. That’s doing work.
Why This Year Feels Different
I’ve been watching technology transform organizations for over 25 years—first on Wall Street, then in government, now consulting with public sector clients. I’ve seen plenty of “this changes everything” moments that didn’t actually change much.
This one feels different. Here’s why.
For the first time, AI isn’t just augmenting what humans do—it’s handling entire workflows. Gartner predicts that nearly half of all enterprise applications will have AI agents embedded by the end of this year. Not as a feature. As a fundamental part of how the software works.
The shift is from AI as a tool you use to AI as a colleague you work alongside.
And the numbers are starting to get real. One company reported their AI agent generated over 2,000 sales leads in a single month. Amazon used AI agents to modernize thousands of legacy software applications—work that would have taken human developers years. Genentech built agent ecosystems to automate complex research workflows so their scientists could focus on actually discovering drugs instead of processing paperwork.
The Part Nobody’s Talking About
Here’s what most coverage of AI agents misses: the technology is maybe 20% of the story.
The other 80% is about people. About changing how work gets done. About the messy, human, organizational stuff that determines whether any new technology actually delivers value or just becomes expensive shelfware.
I learned this the hard way during my years in government. We’d implement new systems that were technically brilliant and watch them fail because nobody thought about how actual humans would actually use them. The technology worked. The transformation didn’t.
The same thing is about to happen with AI agents—at massive scale—unless organizations get smarter about the human side.
What does that look like? A few things:
First, everyone needs basic AI literacy. Not coding. Not data science. Just enough fluency to use these tools, ask good questions, and interpret what they produce. Some researchers suggest a minimum of “30% digital and AI mindset” across the workforce. Whatever the exact number, the principle is right: this can’t be a specialist skill anymore.
Second, someone needs to redesign the actual work. When an AI agent can handle 80% of your customer emails, what do the humans do with that freed-up time? If the answer is “nothing changes,” you’ve wasted the investment. If the answer is “they focus on complex problems and relationship-building,” you’ve actually transformed something.
Third, you need new roles. Job titles that didn’t exist two years ago—AI workflow designers, automation auditors, prompt strategists—are becoming essential. Someone has to guide, supervise, and optimize these systems. The agents are good, but they’re not magic.
What This Means If You’re Not in Tech
I can already hear some readers thinking: “This is interesting, but I work in [government/healthcare/manufacturing/fill in the blank]. This doesn’t apply to me.”
It does. Probably more than you realize.
Some of the most interesting AI agent implementations are happening outside of Silicon Valley. Manufacturing companies using them to process orders. Healthcare organizations using them to handle administrative workflows. Government agencies (slowly, carefully) exploring how to use them for citizen services.
The pattern is the same everywhere: take high-volume, repetitive, rules-based work and let the agents handle it. Free up humans for judgment calls, relationship management, and creative problem-solving.
This isn’t about replacing people. At least not primarily. It’s about letting people do work that actually requires a person.
The Honest Risks
I’d be doing you a disservice if I didn’t mention the concerns.
There’s legitimate debate about whether we’re in an AI bubble—whether the massive investments being made will actually pay off, or whether we’re headed for a correction. The infrastructure challenges are real: systems built for cloud computing don’t automatically work for AI, and security models designed for human workers don’t protect against threats that operate at machine speed.
And the workforce implications are genuinely uncertain. Some jobs will be eliminated. Others will be transformed. New ones will be created. The net effect? Nobody knows yet.
What I do know, from decades of watching technology reshape industries, is that the organizations that adapt thoughtfully—that invest in their people, redesign their work, and approach this with clear eyes—will come out ahead. The ones that either ignore the shift or embrace it recklessly will struggle.
What I’m Watching
Here’s what I’m paying attention to over the coming months:
Multi-agent systems. The next frontier isn’t single AI agents doing tasks—it’s multiple agents working together, passing context to each other, coordinating complex work. This is where things get really interesting (and really complicated).
Industry-specific applications. The generic AI tools are table stakes. The real value will come from agents built for specific industries, specific workflows, specific problems.
The governance gap. Who’s responsible when an AI agent makes a mistake? How do you audit decisions made by software? These questions are largely unanswered, and organizations are flying blind.
The human response. How do people actually feel about working alongside AI colleagues? Early data suggests it’s more positive than you’d expect, but we’re in early innings.
The Bottom Line
The workplace is being quietly, profoundly reorganized. The coworker who handles your routine tasks, processes your data, and keeps work moving forward may not need a desk—or a coffee break, or a salary, or health insurance.
This isn’t coming. It’s here.
The question isn’t whether AI agents will change how your organization works. It’s whether you’ll be ready when they do.
Michael DiNapoli is President & COO of Marcman Solutions. He previously spent 23 years on Wall Street at Citi and Morgan Stanley before leading Florida state agencies including FloridaCommerce and the Florida Development Finance Corporation. He writes about technology, transformation, and leadership.
Connect on LinkedIn.
Leave a Reply