AI’s Next Big Hurdle Isn’t Tech, It’s Your Company’s Structure

According to MIT Technology Review, industry leaders from Everest Group, Concentrix, and Valmont are sounding the alarm on a major roadblock for AI: broken business processes. They describe a widespread “PTSD” of process, technology, skills, and data challenges that cripple organizations with rigid workflows and fragmented information. The consensus is that to unlock AI’s value, companies must stop treating it as a standalone tool and start reengineering entire operational systems. This requires mapping new value streams, designing workflows that blend human verification with AI execution, and building governance and data security from the ground zero. The goal is to shift from simple optimization to complete operational reimagination, moving beyond pilots to create measurable impact.

The Real AI Problem

Here’s the thing everyone’s finally admitting: the tech is the easy part. The real bottleneck, as Shirley Hung from Everest Group puts it, is that most companies are a mess of incompatible systems and rigid workflows. Think about it. You’ve got data locked in a dozen different software tombs, teams that operate in silos, and decision-making that’s slower than a dial-up connection. How is a sophisticated, “agentic” AI supposed to function in that environment? It can’t. The report makes it clear—the central challenge isn’t coding a model; it’s dismantling the organizational concrete that’s been setting for decades.

From Tool to Teammate

So what’s the fix? It’s a fundamental mindset shift. Instead of viewing AI as just a faster calculator or a “virtual worker,” the next unlock is about human-AI collaboration as a system-level capability. Ryan Peterson from Concentrix hits on a key point: humans must stay in the loop to verify content. That means designing workflows where AI handles the heavy lifting of data processing and initial execution, but a human provides the crucial judgment, context, and oversight. It’s not about replacement. It’s about augmentation. This is where you get real acceleration—when AI takes the grunt work off human plates so they can focus on high-impact creation and complex decision-making.

Governance First, Not Last

Now, this all sounds great, but you can’t just flip a switch. Heidi Hough from Valmont offers the crucial, less-sexy advice: expect delays because you need to secure the data. Basically, if you try to bolt this new collaborative system onto a shaky foundation, it will collapse. The smart early adopters are starting with low-risk use cases and building what the article calls “tightly scoped data enclaves.” They’re embedding governance and security into the design from day one. This isn’t an IT project anymore; it requires empowering business leaders to identify where AI can create measurable impact. It’s a full operational blueprint, and the foundation is trustworthiness.

Reimagination Over Optimization

This is where the payoff is. Hung draws the critical distinction: optimization is about doing existing things better, but reimagination is about discovering entirely new things worth doing. That’s the trajectory. The companies that win won’t just be the ones with the best AI models; they’ll be the ones who used AI as the catalyst to tear down their inefficient structures and build something nimbler and more intelligent from the ground up. It’s a daunting task, no doubt. But the alternative is being left behind with a portfolio of cool AI pilots that never actually change the business. The conversation has moved from “Can we build it?” to “Is our company even built to use it?”

For a deeper dive, you can watch the full discussion in the webcast produced by Concentrix.