A new series. Over the coming weeks, we’ll publish one deep-dive on each of the six gaps that decide whether AI becomes a P&L story or a pile of stalled pilots. This is the map.

The conversation changed

Sometime in the last six months, the conversation about AI inside companies changed. It used to be about strategy. Now it’s about results.

CEOs are walking into board meetings with a list of AI pilots and very few numbers on the income statement that show results from these pilots. The dashboards look healthy, but the P&L doesn’t move. Karim Lakhani and his colleagues at Harvard Business School’s Frontier Firm Initiative have named it the last mile problem: enterprises that are “pilot-rich but transformation-poor,” running hundreds of working tools that never converge into a new operating model or bottom-line impact.

The scale of the gap is striking. BCG’s Creating People Advantage 2026 report, drawing on more than 7,100 leaders across 115 countries, found that 90% of CEOs expect AI to reshape their industries — yet only 38% of HR and business leaders rate generative AI as highly relevant to their organization today. That fifty-two point gap between the corner office and the rest of the business is the last mile.

It isn’t one gap. It’s six.

Here’s what a close read of the research — and our own work across CHRO roundtables and direct engagements — surfaced: AI isn’t exposing a single capability gap. It’s exposing six distinct organizational readiness gaps, and not one of them is really about the technology. They’re about how knowledge, data, processes, and people are organized. 

“AI-ready” turns out to be code for a set of organizational conditions that have very little to do with AI itself.

Running beneath all six gaps is a current worth naming up front: fear. Not the science-fiction fear of machine superintelligence — a more pervasive kind. The employee who wonders whether documenting their expertise makes them redundant. The manager who feels unequipped for this new world but can’t say so. FOBO — the Fear of Becoming Obsolete — is the undertow beneath every gap on this list. It’s also why closing them is a people-leadership challenge before it’s a technical one.

And that is exactly why all six gaps converge on one desk: the CHROs.

Why HR owns the last mile

There is no other executive whose portfolio spans all six gaps. Not IT, whose ownership ends at the platform layer. Not the CFO, who sees the outputs but not the operational state behind them. Not the CEO, who has ten functions to integrate and can’t personally inherit this one.

MIT Sloan’s framing isn’t rhetoric: 

HR will be either the architect of AI transformation or the cleanup crew for it. 

The next twelve months will force the choice. Either HR moves into the driver’s seat, or — under board pressure — the CEO finds someone else to drive.

The map: six gaps, six articles

Over the coming weeks, we’ll publish a focused piece on each. Here’s the preview.

1. Knowledge & Process — the hidden cost of undocumented expertise. Roughly 70–80% of how your organization actually works isn’t written down — it lives in the heads of your most experienced people. When AI hits an undocumented step, it stalls, and the bottleneck migrates to exactly the people you can least afford to overwhelm. Worse, asking experts to encode their judgment can feel like documenting themselves out of a job. Coming first.

2. Data — where HR is disproportionately exposed. Ask the CHRO and the CFO, separately, how many people work at the company, and you’ll often get two different numbers. HR data is relationship-based, narrative-heavy, and never had the audit discipline that makes financial data machine-readable. It was built for humans to interpret, not for systems to reason over — and AI surfaces that gap fast.

3. Coordination — the battle to be the front door. Every enterprise tool now has an AI assistant tab and a claim to be where work starts. The result isn’t one front door; it’s a hallway full of them, none agreeing on what “context” or “user” even means. The result is uncoordinated AI agents that deliver underwhelming results and a poor employee experience. 

4. Capacity — when AI activation is up but results are flat. AI license number utilization looks great; outcomes don’t move. “Workslop,” fatigue, and shadow AI spread without shared norms, and an AI push can raise anxiety faster than it raises productivity. Capacity is a problem of time, training, trust, and psychological safety.

5. Governance — the infrastructure of trust. Good governance goes well beyond compliance. Employees collaborate with AI they believe is being used fairly; when that trust breaks, the damage reaches every other gap. Decision rights, an explainability standard, and a real audit cycle — plus a sharpening regulatory backdrop — make adoption sustainable rather than merely fast.

6. The Manager Gap — redefining what it means to lead. The job of management is being structurally redefined, and almost no organization has told its managers. What it means to set direction, delegate, and develop talent looks different when part of the team is software. Most managers are experiencing this as exhaustion; it’s actually an organizational-design problem — and the fulcrum the other five turn on.

Where to start

There’s no single plan that closes all six at once, and the most effective leaders aren’t trying to. They pick one workflow, one team, one month — a deliberate experiment that exposes every gap at small scale — and use what they learn to redesign the work and sequence what comes next.

If you’d like a clearer picture of where your function stands, our Six-Gap Readiness Scan benchmarks you across all six dimensions in three weeks and tells you exactly where to begin. Often a 30-minute conversation is enough to know whether a readiness scan or a focused adoption pilot is the right first step.

Reach out at people-ai-hr.com/contact-us.

The architects of the last mile are being chosen right now. The cleanup crews will be obvious later.

Next in the series: Gap 1 — Knowledge & Process.