There’s a pattern repeating itself across enterprise technology right now, and it has nothing to do with which AI platform you’ve licensed.
Leadership announces a transformation initiative. Headcount gets cut in the name of efficiency. Automation tools get deployed. Output metrics improve … briefly. Then something quietly breaks in the middle layer of the system that nobody can quite name, and twelve months later the organization is having the same conversation again, only now the tools cost more and the practitioners who knew where the bodies were buried are gone.
This isn’t an AI problem. It’s a structural problem that AI is making more expensive.
The wrong question
Most AI transformation conversations start with “how do we automate this?” That’s the operational question. It’s the right question inside the wrong frame.
The frame is the org chart. The line-and-box hierarchy was designed for a pre-AI world … to manage repetitive manual labor, fixed headcount costs, and sequential coordination. Every AI initiative that gets bolted onto that structure inherits its assumptions: work flows down reporting lines, accountability lives in positions, and the measure of success is output per headcount.
Jerome Kanter wrote about this in 1967. His book The Computer and the Executive mapped exactly how organizations would mismanage the introduction of computing power by trying to absorb it into existing hierarchy rather than redesigning around it. Substitute “AI” for “Computer” throughout. The text reads as contemporary.
We’re repeating a sixty-year-old mistake with a compressed timeline and higher stakes.
What’s actually accumulating
When an organization traps its expert practitioners in shallow, repetitive tasks … the work that AI should be absorbing … it creates what I call Management Debt.
Management Debt is the accumulated cost of organizing around position and headcount rather than outcomes. It compounds. The practitioner who spends three years running regression scripts that an AI agent could execute is not building the domain knowledge that would let them govern those agents effectively. The apprentice who should be learning from that practitioner isn’t getting the apprenticeship. The organization loses the ability to evaluate what it’s losing.
This is the Craft Death Spiral: AI redundancy washing → apprentice pipeline collapse → craft knowledge death → the organization can no longer evaluate its own outputs.
The Wiles, Hsu, Bedard, and Kropp study (BCG/Boston University, 2026) documented the governance version of this in real time. When organizations encode AI into their hierarchy … giving agents names, job titles, org chart positions … managers reduce error-detection rates by 16% and escalation requests increase by 44%. The human stops owning the output because the frame says someone else does. The org chart isn’t just misaligned with AI. It actively degrades the oversight that AI-augmented work requires.
The Bed Bath & Beyond inversion
There’s a useful case study in an unlikely place.
Bed Bath & Beyond spent years trying to bolt digital capability onto a retail structure. Web presence layered on top of stores, digital team alongside physical team, two structures competing for the same customer. It didn’t work. The company eventually burned down the retail structure, rebuilt around digital as the core, and then … this is the part most people miss … selectively reintroduced physical presence in targeted areas where it created genuine, measurable value.
The expertise … merchandising, category knowledge, supplier relationships … was preserved. The structure was rebuilt around it.
That’s the inversion most AI transformations refuse to make. The expertise your practitioners carry is the asset. The org chart position they occupy is the liability. The organizations winning right now aren’t automating their old model faster. They’re designing around a new one with their best people at the center of it.
High-Yield Governance
The alternative to Management Debt is a Strategic Portfolio model … what I call High-Yield Governance.
The logic is straightforward, and it maps deliberately to the language boards and CFOs already speak.
AI handles the repetitive layer. Regression execution, documentation generation, routing, pattern-matching. This is the “interest” earned on the manual labor investment … it gets reinvested, not pocketed.
Practitioners become Liquid Capital. Instead of being frozen in rigid hierarchy, expert practitioners flow to the most critical needs as they arise. They’re not “QA testers.” They’re Asset-Backed Resources … appreciating in value because they’re doing the work that compounds: root cause analysis, exploratory risk discovery, AI validation, documentation archaeology.
The leader becomes the Architect of Solvency. Not the manager of headcount. The designer of the operating model … the person who knows which human judgment points are non-negotiable, who can read the deep technical stack, and who can demonstrate to the board that every action taken by the human-AI partnership is driving measurable appreciation of the company’s technical assets.
The Deep Stack Dividends this model generates aren’t abstractions. They’re the categories of high-value work that have always been needed and have always been uneconomical to fund at scale under the old model:
- Documentation Archaeology … recovering buried system knowledge before the people who carry it leave
- Root Cause Analysis as a Service … finally funding the defect categories that preventative testing never reaches
- AI Validation … maintaining craft-level oversight of AI outputs (testing the tester)
- Exploratory Augmentation … deep-stack risk discovery at a scale solo practitioners could never achieve manually
Under the old model, these categories lose the budget conversation every quarter. Under High-Yield Governance, they’re the yield. A CFO will fight to preserve a dividend-generating asset. They will not fight to preserve a cost-center line item.
The diagnostic question
Not every leader is ready for this conversation. The ones who are will pause on a simple question:
When your board or CEO asks what you’re doing with AI … are you showing them a headcount reduction or a capability strategy?
The leaders who pause already feel the gap. They’re being asked to demonstrate transformation while executing what looks increasingly like dismantlement. They need vocabulary and a framework to close that gap … and to be the person in the room who has the answer when peers are still explaining why the last initiative underperformed.
The leaders who answer without pausing … “headcount reduction, absolutely” … are optimizing the old model. They may get there eventually. They’re not ready yet.
The shift nobody wants to name
There’s a political dimension to this that’s worth naming directly.
The move from org chart to work chart … from positional hierarchy to outcome network … is a shift from position power to knowledge power. For leaders who have spent decades accumulating positional authority, that shift feels like a threat. In some cases it is one.
But here’s the honest read: position power is already eroding whether they adopt this model or not. AI is performing the coordination, information-filtering, and routine execution that historically justified middle management layers. The Great Flattening is not a prediction. It’s a present-tense trend documented in Microsoft’s Work Trend Index, visible in five consecutive quarters of profitable-quarter layoffs at major tech companies, and now measured empirically in oversight quality degradation.
The question isn’t whether the structure changes. It’s whether the leader is holding the pole when it does.
The Architect of Solvency role carries more durable authority than any org chart position because it’s knowledge power that compounds. The person who designed the new model, who understands which human judgment points are non-negotiable, who can articulate the difference between Technical Solvency and Optical Integrity to a board … that person doesn’t get flattened. They become the fixed point the flattening reorganizes around.
The bottom line
We’re not asking for a budget to do more testing.
We’re presenting a Capital Allocation Strategy that uses AI to unlock the hidden equity in existing systems. We’re moving from Technical Bankruptcy … deep risks unaddressed, practitioners trapped in shallow work, AI bolted onto a structure that no longer fits … to High-Yield Solvency, where human expertise compounds, AI generates measurable returns on manual labor, and the organization is structured around outcomes rather than reporting lines.
The companies that capture disproportionate value in the next cycle won’t be those that automated fastest. They’ll be those that knew what not to automate … and built governance that made that distinction visible to the board.
Appreciate, Don’t Liquidate.
Rick Cavallaro is President of RJCadvisors Inc. and the author of The New Beautiful Testing. For the full framework and white paper, visit qameditations.com.