Why intelligent solutions are becoming the operating system of every serious business

The tough questions in the boardroom have shifted. A year ago, leaders were debating when and how to invest in AI and agents. Today the question is sharper: why hasn’t the spend started paying for itself?

That move from “let’s act” to “prove it worked” is where a lot of organizations get stuck. It’s an uncomfortable place to sit — but the distance between investment and return is bridgeable. The bridge is design.

I’ve spent more than twenty years in this field. I watched the cloud arrive and listened to every argument for why it would never stick. The companies that dug in their heels are the ones carrying the technical debt today. Agentic AI is the same story, with the same resistance and, for those who wait, the same ending. The difference is that the window is wide open right now in a way I haven’t seen in years.

Everyone is spending. The returns haven’t caught up.

The data all points the same direction. Last year 85% of organizations increased their AI investment, yet only 6% could show a measurable return inside twelve months.¹ Some 88% now use AI in at least one function and 62% are already trialing agents — but fewer than 40% report any enterprise-level financial impact, and roughly two-thirds have never taken AI past the pilot stage.²

The appetite is real and so is the budget. What’s usually missing is the strategy underneath it.

I recently sat with a team that had built an agent to automate a single task. The accuracy was genuinely good. Then someone asked the obvious question: how often does this process actually run? Once a week. About thirty minutes of manual effort each time. A few dollars of someone’s time. The cost to build, run, and maintain the agent dwarfed all of it. Technically sound, commercially a loss.

That’s a bit like hiring a person and then working out their job afterward. You’d never do that. You define the role, agree what good looks like, and set the budget before day one. Skip that with an agent and you get exactly what the numbers above describe: activity with no payoff.

The real problem isn’t the technology

Agents are new. The fundamentals aren’t. Data, process, and people have decided the outcome of every technology rollout I’ve ever worked on, and that hasn’t changed. Break any one of the three and the rest breaks with it. The model you run on top makes no difference — garbage in, garbage out still holds.

When an AI initiative stalls, it’s rarely a technology failure. It’s a foundation failure. The data isn’t aligned to the use case. The process is too immature to hand to an agent because it lives in someone’s head rather than in a documented, repeatable workflow. And governance gets treated as a clean-up job for after launch. You can’t build a reliable digital coworker on a foundation you never designed.

Results across industries — what a real foundation produces:

Five reasons agentic AI investments miss

Most implementations don’t fail because the technology doesn’t work. They fail because the foundation was never designed to carry them. Here is what I see going wrong, again and again, in organizations of every size.

1. The wrong use case. Most teams start with what looks exciting rather than what has a defined return. Before building anything, answer three questions: what’s the expected output, what does success look like, and where’s the break-even point?
2. Data that isn’t ready. The most capable agent on the market can’t paper over data that is misaligned, incomplete, or untrustworthy.
3. A process too immature to copy. An agent only does what it’s designed to do. If the workflow lives in someone’s head instead of in documented, governed logic, it can’t be automated reliably.
4. An agent built beside the work instead of inside it. An agent that sits next to a process is an optional tool. An agent embedded in the process is how the work gets done. Successful implementations are almost always the ones designed into the middle of how people actually operate.
5. Underestimating the human side. Adoption kills more AI projects than bad technology ever will. Buying a tool isn’t the same as embedding one. The teams that get this right treat change management, training, and process redesign as part of the build — not a follow-up.

The shift that matters most

I describe this moment as a move from the applications of the past to the intelligent solutions of the future. For decades our job was to walk into a business, learn how it worked, and reproduce that in software: create a sales order this way, approve a purchase order that way. The system recorded it, structured it, reported on it. That was valuable. It isn’t a differentiator anymore — a $19 accounting license does the same thing.

The operating system of every business from here

The layer where real value now sits is what I call intelligent solutions: AI-infused versions of business processes that analyze, synthesize, recommend, and act. They move an organization from human by default — where a person has to start, approve, and close everything — to human by exception, where the system handles what it can and only escalates the calls that genuinely need a human. This is the operating system competitive organizations will run on.

The opportunity hidden inside the paradox

Most leaders read that 6% figure as a warning. I read it as a differentiator, and a powerful one. Only around 10% of organizations have agents scaled in any single function today.² The field is wide open. The leaders who lay the right foundation now — the ones who treat intelligent solutions as the new operating system rather than the next experiment — will be very hard to catch.

The way through is to be an early adopter, but a controlled and structured one. That means defining outcomes before you build, designing governance in from the start, and choosing a partner with the industry knowledge, platform depth, and delivery record to move you from idea to operating reality — not from idea to impressive demo. The intelligent solution, the digital coworker, the operating system of every serious business: that’s what’s being built today.

Design Your Results

Ready to move from agentic AI potential to measurable payoff? HSO helps you design the path — from first agent to enterprise outcome. Let’s talk.

→ Explore Agentic AI, by Design

Sources

¹ Deloitte, AI ROI: The Paradox of Rising Investment and Elusive Returns, October 2025. Survey of 1,854 executives across 14 countries in Europe and the Middle East.

² McKinsey & Company, The State of AI 2023–2024 and McKinsey Global Survey on AI.

³ Gartner, Predicts 2024–2026: AI Strategy and Execution.

The post The AI ROI Paradox Is Real. And It’s Your Biggest Opportunity. appeared first on CRM Software Blog | Dynamics 365.