Across industries, companies are moving from AI pilots to real deployments and seeing measurable results. Teams are generating insights, completing research, and drafting work in a fraction of the time. What took months can now take days — what took days can now take minutes.

But then, momentum slows.

That’s the moment companies learn that productivity gains alone don’t automatically translate into business transformation.

McKinsey finds that 62% of organizations are experimenting with AI agents, but only 39% report an earnings impact. Gartner expects more than 40% of agentic AI projects to be canceled by the end of 2027.

The reason is simple: one critical part of the workflow hasn’t changed. Organizations are still relying on people to bridge the gaps between systems.

People as middleware 

For decades, enterprise workflows were designed around people moving information between applications. Every piece of software — a CRM, an ERP, a project management platform, a productivity suite — is built for one type of user: a person who manually stitches a process together because the applications themselves don’t fully connect. 

People swivel-chair between them, copying information from emails, clicking through numerous interfaces and other tools. I call this the “toggle tax.” Researchers writing in Harvard Business Review found that employees switch between applications roughly 1,200 times per day, consuming nearly 9% of the workweek in reorientation and context switching alone. 

Rethinking the role of SaaS 

It’s a hefty, time-intensive tax, and agents are uniquely positioned to eliminate it. That promise has sparked predictions of a “SaaSpocalypse.”

The argument is straightforward: If agents can access the same data, the obvious question is: why do you need the app at all?

I think the better question is what the app becomes.

In an agent-driven world, SaaS will no longer be what people log in, navigate, and toggle. It evolves into a system that provides agents with the unique context they need to act reliably and securely. Their value was never the interface — it has always been the data, relationships, governance, business rules and logic that lie underneath. 

As agents coordinate work across applications, people can spend less time moving information and more time serving customers, making decisions and driving business outcomes. By 2028, Gartner predicts that 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024. 

Redesigning work for the agent era

Realizing the opportunity ahead requires more than deploying new technology. Microsoft’s latest Work Trend Indexrevealed that only one in four employees see leadership alignment on AI. The productivity J-curve explains why that matters: transformative technologies deliver their greatest gains only after organizations align around a common vision and redesign the work processes the technology supports.

Leaders don’t need to reinvent the business overnight to move forward; in many cases, the next step is simpler than they think. Start with a workflow that cuts across multiple applications and has clear business value. Improve it, learn from it, then scale. As momentum builds, identify where the toggle tax is still slowing operations down and simplify those workflows as well. 

The more capable agents become, the more they’ll depend on connected enterprise applications. But their greatest impact comes when organizations redesign workflows around this new way of working. As that evolution takes hold, people step out of the role of human glue. Instead of coordinating across systems, they focus on what only humans can do: make decisions, build relationships, and move the business forward.  

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

This story was originally featured on Fortune.com