Where AI Meets HR: Q3 2026

Over the third quarter of 2026 we sat down with CEOs and Chief Officers across the Fortune 500 and the world's most recognized companies, spanning packaged foods, financial media, pharmaceuticals, hospital systems, semiconductors, public education, facilities services, and asset management. What follows is a summary and analysis of the trends spanning those conversations.

Trend #1: AI and Revenue: Is This Moving the Business, or Just Keeping Us Busy?

AI and revenue is the question sitting underneath everything else. The shift this year is from adoption to proof: leaders have stopped asking whether to adopt AI and started asking whether the work already underway is producing anything, or whether it is elaborate motion. A year ago the question was permission. Now it is evidence, and the uncomfortable part is how few can produce any.

A Chief Officer at a packaged-foods company came to our conversation straight from an internal meeting on mapping AI initiatives to revenue growth. The framing was precise: personal productivity and supply chain are table stakes now, and the open frontier is revenue and category management. The real question is whether any of this touches the top line.

At a financial-media organization the same question sits unanswered despite enterprise-wide tool access and custom builds across advertising, marketing and editorial. Considerable activity, genuine capability, and still no answer as to whether AI can power the future business model rather than merely trim workflow cost.

A spirits company put it more bluntly. Peers are barely discussing AI implications at all, and the industry is early in a disruption with years still to run.

The skepticism is sharpening rather than softening. A semiconductor manufacturer is pushing AI across every function with one demand attached: first-principles rethinking, not AI bolted onto existing process. Leaders have learned to recognize AI theater, and they are increasingly unwilling to fund it.

What none of them have is proof. The gap is not awareness, it is evidence. They want a case study with real profit-and-loss impact they can put in front of a chief financial officer, and that evidence does not yet exist at any scale.

Trend #2: Employee Fear: The Anxiety Nobody Puts on the Agenda

Employee fear is the real constraint on adoption, and it is almost never named as such. Job displacement anxiety came up in nearly every conversation, arriving sideways as a change management challenge, a communications problem, a culture issue. Underneath, the concern is identical across sectors: employees do not believe leadership when leadership says AI will not cost them their jobs.

A Chief Officer at an academic medical center named it directly. The fear is that AI investment leads to job cuts regardless of what leadership says. The response there is deliberate slowness, refusing to overinvest early precisely because overspending would later force the cost-cutting that proves the fear correct.

A public school district reached the same conclusion from a different direction. The framing has to be augmentation and human-AI collaboration, not replacement, and getting it wrong is expensive in ways that are not only financial.

The most sophisticated version came from a pharmaceutical company, where the concern is not today's roles but the feeder pools, the entry and mid-level positions where future senior leaders are actually made. If AI absorbs that work, the pipeline that produces executives quietly stops functioning, and nobody notices for five years.

Fear is not evenly distributed, and it does not follow the demographics people expect. It concentrates in administrative and operational staff rather than clinical or research teams, and in rule-based roles generally. At a food-ingredients company, some of the most enthusiastic adopters on the talent team are the oldest people in the room.

One incident at a senior-living operator illustrates what leaders are actually afraid of. A manager used a general-purpose AI tool to draft documentation of a serious workplace incident and increased the organization's legal exposure in doing so. Stories like that travel internally faster than any enablement program.

What leaders are asking for is not reassurance. It is a narrative specific enough to their own organization that their people will actually believe it. Generic augmentation messaging has stopped working.

Trend #3: Measurement: I Cannot Prove Any of This Is Working

Measurement is where AI investment currently breaks down. Spending has outrun the ability to measure it, so money is going out and almost nobody can yet show what it bought. The problem surfaced constantly, and most sharply among people leaders under pressure to justify their function in business terms.

At a facilities-services company, reporting is still done by hand: weekly updates entered manually, business-intelligence access described as restricted and painful, applicant data that hiring managers do not keep current. What they want is not advanced analytics. It is a live view of time-to-fill, the applicant funnel, salary comparisons across markets, and source attribution.

That pattern repeated nearly everywhere. No live pipeline visibility, because tracking data is stale or locked away. No aggregation, with data stranded across systems nobody has connected. No source attribution. No before-and-after framework for the AI training itself. And no succession visibility, because the org chart is a static document rather than a living talent inventory.

A technology company with genuinely strong HR infrastructure still has no strategic AI roadmap, and flagged unstructured AI spend as a live risk. Teams accumulate real cost building what one leader called glorified spreadsheets, with no framework for evaluating whether any of it worked.

The ask is smaller than it sounds. Almost nobody is requesting sophisticated modeling. They want basic visibility that does not currently exist, connected clearly enough to business outcomes that it can be shown upward.

Trend #4: Transformation Fatigue: In the Middle of Something Enormous

Transformation fatigue explains most of what looks like resistance. What stalls AI work is capacity, not conviction: a significant number of leaders who seem slow to move are not resistant, they are saturated. Acquisitions, restructuring, platform migrations and leadership transitions are consuming exactly the capacity that AI investment would require, and reading that as reluctance gets the situation backwards.

A semiconductor manufacturer is mid-acquisition with a long integration ahead and executive hiring frozen. A biotech closed an acquisition during the period we spoke, with contracts frozen and an outsourced AI governance policy left unfinished.

A public health system is rolling out a new performance platform, implementing compensation changes and fielding an engagement survey at the same time. A wealth management firm hired well beyond its usual annual pace through acquisitions and is now deliberately paused. A professional association is mid-transition between chief executives, with every prior priority being reorganized.

Acquisition activity across this group is best described as intense. The consequence is a recurring pattern: decisions freeze, timelines stretch, and the window to act is both narrow and unpredictable. Being in place before a deal closes is a categorically different position from trying to establish a relationship during integration.

None of this is a reason to disengage. It is a reason to reposition. The leaders carrying the most complexity are precisely the ones who need something that reduces load rather than adding to it.

Trend #5: The Manager Gap: Where It Actually Breaks

The manager gap is where AI strategy actually breaks, and it is not at the executive tier. Leaders open with strategy, culture and transformation, but press on where the breakdown occurs and the answer is almost always middle management. Managers are the transmission between leadership intent and employee experience, and in most of these organizations that transmission is slipping.

A public health system running a round-the-clock operation confirmed that its internal data matches a specific and unflattering pattern. Accidental managers promoted for technical performance. Conflict avoidance. Cohesion gaps across shifts. Change fatigue. Burnout concentrated at the manager tier. An inability to influence across silos.

At a food-ingredients company, AI adoption across the talent team varies wildly. The variance is not generational and it is not about tooling. It is capability and confidence. Meanwhile the expectation being set for that function, covering workforce planning, proactive pipelines and genuine strategic partnership, assumes a level most of the team has not reached.

A healthcare practice group recently assessed all of its practice managers and is concentrating its entire development effort on the underperforming sites rather than on average or high performers. That focus is itself a finding. The manager gap is not evenly spread, and the highest-leverage work sits at the bottom of the distribution.

A peer-advisory membership organization pointed at the same tier from another angle. With members joining and leaving continuously, manager development there cannot be an event. It has to be a standing capability.

Three forces converge on this layer. Managers must model AI use for their teams. They translate leadership messaging into behavior. And they are the primary driver of voluntary turnover. Very few organizations invest in them anywhere near proportionally to that load.

Trend #6: Governance Friction: I Want to Move, but I Cannot Move Fast

Governance friction, not appetite, sets the pace of AI adoption. Compliance, procurement and legal review are the structural brake on nearly every leader who wants to act. This is not a footnote to the other five trends. It is frequently the binding constraint.

At an investment bank, every external engagement runs through review and communications approval before any commitment. At an academic medical center, implementation waits on board approval by design, with governance structures targeted for completion before use-case evaluation begins. Neither is resistance. Both are institutional caution working as intended.

At a public health system, vendor approval is a prerequisite to any contract, and an intellectual-property clause concerning AI once caused a long delay with a prior supplier. That approval is not a formality. It is a real timeline risk, and also a real asset once cleared.

The third form is neither regulatory nor procedural. At a mortgage lender, IT is blocking an assistant rollout despite an enterprise license providing the relevant data protections, and despite employees having asked for AI training. The stated concern is data security; the pattern suggests something broader about control and accountability. At a family-owned distributor, ownership dynamics produce unfocused meetings where AI is barely raised at all, which is a governance vacuum rather than an obstacle.

The organizations making progress share one characteristic. There is a clear internal champion who owns the AI agenda and holds either the authority or the relationships to move it through. Where that person does not exist, intent does not convert into anything.

The practical implication is to reduce friction rather than price. Clean paperwork, clear intellectual-property terms and a process that fits inside existing procurement is worth more than a discount.

What It Adds Up To

Six forces, and they compound rather than sit side by side. The revenue question cannot be answered without measurement infrastructure. Measurement cannot be built by an organization saturated with integration work. Employee fear does not resolve without managers capable of holding the conversation. And none of it moves without a champion who can clear governance.

The useful response is narrower than it first appears. Produce evidence of profit-and-loss impact rather than frameworks. Hand leaders a reframing narrative specific enough for their own people. Build the basic dashboard that does not exist. Position against work already in motion instead of adding to it. Lead with the manager tier rather than the C-suite. And make saying yes procedurally easy.

One caveat about this data. It reflects a single month of conversations. The consistency between sectors is striking, but a month is a short window, and companies willing to take this meeting are not a random sample of the market.