top of page
Search

June 1, 2026: Most Organizations Are Running AI Experiments. Almost None Are Capturing the Returns.

  • Writer: James Sale
    James Sale
  • Jun 1
  • 4 min read

Organizations are experimenting with AI at scale. They are not, in most cases, getting much back for it.


Per McKinsey's State of Organizations 2026 report, 88% of organizations are experimenting with AI, but 81% report no meaningful bottom-line gains. Meanwhile, 86% of leaders say their organizations are not very prepared to adopt AI in day-to-day operations. These are not numbers that suggest a technology adoption problem. They suggest an organizational design problem.


That distinction matters, because the solution looks completely different depending on which problem you're solving.


Most HR Functions Are Still in the Early Innings

Phenom's State of AI & Automation for HR: 2026 Benchmarks Report, which analyzed nearly 500 companies, found that 83% demonstrated low AI and automation maturity. Across those companies, most are not at the workflow redesign stage. They're still figuring out what tools they have and whether anyone is using them consistently.


There are pockets of real progress. Per the Phenom report, 90% of healthcare organizations are using automated candidate campaigns, and 68% of financial services firms are using AI for candidate matching. Those are meaningful numbers. They also represent a fairly narrow slice of what AI can do inside a talent function, and they are concentrated in industries with volume hiring pressure and clear ROI cases.


If you lead HR at a company outside those sectors, the data says your function is probably still in early-stage testing. The honest question to ask your team: are your current pilots building toward a repeatable workflow, or are they building toward a demonstration?


> Worth doing now: Map one high-volume, low-variance HR process (candidate screening, onboarding document prep, interview scheduling) and ask whether there is a clear owner accountable for operationalizing AI in that specific workflow by Q3.


Marketing Hit Operational Maturity First, and the Lessons Apply Everywhere

Jasper's State of AI in Marketing 2026 puts a fine point on where the industry has moved: "The experiment is over and the operational era has begun." Marketing is not unique in its technology access. It is ahead of most functions in treating AI as an operational tool rather than a pilot project.


That shift involves three things the report examines directly: governance (who approves AI-generated outputs and under what conditions), role changes (which responsibilities shift when content production accelerates), and measurement (how you hold AI-assisted work accountable to outcomes, not just output volume).


Those three levers apply in every function. If your team is still evaluating AI tools, marketing's trajectory is worth studying. The transition from pilot to operations is not primarily a technology question. It is a people and process question.


Manager-Level Adoption Is Climbing, But Tool Governance Is Lagging

Beautiful.ai's survey, published recently, shows daily AI usage among managers nearly doubled, rising from 18% to 34%. 77% say they adopt AI for efficiency and productivity. The number that deserves more attention: only 53% are using employer-approved tools.


Nearly half of managers using AI daily are doing it outside any governance framework their organization controls. That is not a compliance footnote. It means your team's work product is potentially running through tools your legal, security, and IT teams have not vetted. It also means your organization has no visibility into what workflows have quietly been redesigned around AI.


The instinct to lock everything down is understandable, but it usually backfires. The managers most actively using unapproved tools are often your highest performers looking for leverage. The better move is to catch up to where your people already are, build clear approved tool lists, and create a fast path for managers to request vetting of new tools rather than just banning what you haven't reviewed.


The Double Transformation Problem Every C-Suite Needs to Name

McKinsey's framing from their State of Organizations report deserves to sit in every leadership team's Q3 planning conversation. The report calls for a "double transformation": technical and organizational. That is a diplomatic way of saying that buying and deploying AI is the easy part. Redesigning workflows, reallocating talent, and changing how decisions get made is the hard part, and most organizations haven't started it seriously.


86% of leaders reporting low preparedness for day-to-day AI operations while 88% are actively experimenting is not a contradiction. It is a description of what happens when technology adoption runs ahead of organizational readiness. Tools get deployed into workflows that were never designed to use them. Adoption numbers go up. Impact numbers stay flat.


Grant Thornton's 2026 AI Impact Survey, which surveyed 950 C-suite and senior leaders, examines this directly. The report analyzes what Grant Thornton calls the "AI proof gap" and what separates firms that capture real ROI from those that remain stuck in demonstration mode. The consistent differentiator, per the report, is that leading organizations pair technology deployment with explicit workflow redesign and accountability structures, not just access.


If your organization is in the majority on the McKinsey numbers, the path forward is not more pilots. It is naming which three workflows will be fully redesigned before year-end, who owns that redesign, and what the accountability metric is.


The gap between "88% experimenting" and "81% seeing no returns" does not close with more experimentation. It closes with organizational decisions that most leadership teams have been putting off.


If you want to stay current on how AI is actually changing how organizations operate, where the returns are real and where they're still theoretical, Agenticism is where those stories land. Practical, grounded, written for professionals making real decisions.


Sources

Recent Posts

See All
bottom of page