The Fallacy of Productivity Gains (and Why Agentic AI Won’t Save Us)

Allow me to vent a little. I know I am in a privileged position, and I love my work as an industry analyst, but in virtually every briefing I sit through and each conference panel I endure, someone stands up to tout their software’s ability to radically improve productivity. Forgive the sarcasm, but if even a fraction of these claims were true, knowledge workers would be operating at 500% of the efficiency we managed just a few years ago. We are not, clearly and obviously not, and there is no statistical evidence to suggest that we are.

To be clear, productivity does improve. But detailed, longitudinal studies consistently place that improvement in the 1.5%–2% per annum range, hardly the stuff of revolution and transformation. Economists will argue until sunset about the nuances of these measurements, and they rarely reach a satisfactory conclusion. But, in my opinion, most of the hyperbolic claims I hear from software vendors are built on sand. So why does this persist? Why do buyers nod along, and why do tech vendors double down decade after decade?

One of the few benefits of having been around the block a few times is the painful recognition that lessons learned in one era are rapidly forgotten in the next. Take the late-1990s craze for Six Sigma and LEAN. The core lesson then was simple: When you push for more output in less time without addressing the underlying process, quality suffers. Quick wins in productivity often lead to slower outcomes downstream, buried in rework and customer churn. As the old adage goes, “You can have it fast, cheap, or good—pick two.”

The Arrival of Agentic AI

Which brings me, rather inevitably, to the topic of agentic AI. Today’s pitch is that autonomous agents—systems that plan, execute, and iterate on behalf of the knowledge worker—will finally deliver the productivity nirvana that CRM, ERP, hyper-automation, and RPA promised but failed to keep. The sales decks are glossy, and the demos are often super slick. But I fear we risk simply repeating the same ethical and operational failures of years past.

Let’s talk about the positive correlation: the “lean” scenario, where higher productivity actively drives higher quality. This is the reality in well-run organizations. They improve the process to eliminate waste, rework, waiting, defects, and unnecessary handoffs. Doing things right the first time is, in fact, faster and more productive. In this world, a well-deployed agentic AI could be a “force multiplier,” automatically routing exceptions or validating data at the edge. But here is the uncomfortable truth most vendors won’t utter: The real correlation is not between productivity and quality. It is between “management quality” and productivity.