The News
A PR pitch on behalf of Michael Privat, an engineering leader overseeing 500-plus engineers at a healthcare IT company that processes data for more than half of all U.S. health claims, presents an argument about the economics of AI-driven productivity. Privat’s central claim is that AI has not underdelivered on efficiency but that most organizations have directed those gains at pre-existing, suboptimal processes rather than eliminating them. The pitch cites a Deloitte survey of 3,235 business leaders finding that 74% want AI to drive revenue growth while only 20% are actually seeing it, framing the gap as a deployment and process discipline problem, not a technology problem.
Analyst Take
The Deloitte number is striking, but it is not surprising to anyone who has watched enterprise AI rollouts closely. The productivity paradox Privat describes has a structural cause that most ROI frameworks miss: organizations are measuring output, not outcomes. More code shipped, more tickets closed, more queries resolved. None of that translates to revenue or margin if the underlying process being accelerated had no business existing in the first place.
The Efficiency Trap Is Real, and the Data Backs It Up
Privat’s observation that “work is like gas” has a quantitative analog in how engineering time is actually allocated today. ECI Research’s 2026 Application Development survey found that 65.2% of respondents said only 0–20% of engineering time is spent on net-new innovation. That is a striking figure. If the majority of an engineering organization’s capacity is consumed by maintenance, integration debt, and operational toil, then an AI tool that makes those activities 30% faster is not a strategic win. It is an efficiency gain that compounds the wrong priorities. The AI amplifier, as Privat frames it, faithfully reproduces whatever it is pointed at.
This is precisely the mechanism that explains the Deloitte gap. Revenue growth requires innovation capacity. When AI absorbs into the existing maintenance burden rather than freeing engineers to work on differentiated product, the ROI lands in cost avoidance at best, nowhere at worst. CFOs cannot find it because it was never allocated to a revenue-generating activity.
Why the Headcount Math Was Wrong from the Start
The framing that dominated AI’s commercial pitch in 2023 and 2024 was essentially labor substitution: fewer engineers doing the same work. Privat’s argument implicitly rejects this as the wrong unit of analysis. The question is not how many engineers you need to maintain your current processes. It is whether your current processes deserve to survive at all.
Healthcare IT is a particularly pointed context for this argument. The sector carries some of the densest regulatory and interoperability obligations in enterprise software, which means that legacy process proliferation is not just cultural inertia but often a compliance artifact. At a company processing the majority of U.S. health claims, the operational surface area is enormous and the tolerance for disruption is low. That makes it harder to delete the bureaucracy AI is now accelerating, and it makes the organizational discipline Privat is describing genuinely difficult to execute.
What This Means for ITDMs and Developers Alike
For IT decision-makers, the implication is direct: AI tool procurement without process rationalization first is a budget line that will not close. Buying a coding assistant for an engineering team that spends the bulk of its time on maintenance work does not structurally change what that team produces. ECI Research’s 2026 Application Development survey found that only 3.6% of respondents said 41–60% of engineering time is spent on net-new innovation, confirming that the innovation capacity problem is widespread, not isolated to healthcare or any single vertical.
For developers, the argument is more nuanced. AI-assisted tooling is genuinely accelerating inner-loop cycles, and the productivity lift at the individual contributor level is real. The problem is organizational, not technical. The friction sits at the boundary between what an individual engineer can now do faster and what the organization’s process architecture allows to count as value. Closing that gap requires leadership decisions, not better prompts.
Looking Ahead
The conversation Privat is trying to start will intensify over the next two to four quarters as enterprise AI budgets come up for renewal and CFOs demand clearer attribution. The organizations that emerge with defensible AI ROI stories will be those that treated the productivity gain as a forcing function for process elimination, not a reason to tolerate more complexity. The ones that cannot tell that story will face a reckoning, particularly in capital-intensive verticals like healthcare IT where the cost of maintaining status quo processes is already high.
Longer term, the market is moving toward a clearer bifurcation. Vendors and consultants who can help organizations identify which processes should be deleted before AI is applied will command a premium. The pure-play productivity tooling market, by contrast, will commoditize faster than most pricing models currently assume. Privat’s argument is a preview of the organizational design conversation that enterprise technology buyers need to be having right now, and are not.
Stay Ahead of Application Development Trends
Get weekly analyst insights, research notes, event coverage, and AppDevANGLE updates delivered directly to your inbox.
Subscribe for Weekly Insights
Join technology leaders, practitioners, and GTM teams following the trends shaping modern software delivery.
Looking for deeper research access?
Explore ECI Research reports, survey insights, and market analysis through the ECI Research Portal.
