Why better matters.

Field research at Walmart, Amazon, Ericsson, Ramp, and Medtronic. Companies that redesigned cross-functional decisions around human-AI collaboration, with people holding context and the final call, moved faster and decided better.

AI use jumped to 61% of service firms and 51% of manufacturers. 4% of service firms and no manufacturers laid anyone off because of it. Retraining is how most firms are adjusting.

49% of customers say they would use a company's chatbot. 7% actually did in their last service interaction. After one bad experience, 27% would try again. The experience decides adoption.

Three of four AI high performers fundamentally redesigned workflows this year. One in four of everyone else did. The share of companies seeing any profit impact from AI stayed flat at 37%.

Three in four leaders say humans collaborating with AI agents create more value than automation on its own. Only one in five says their organization is ready to redesign processes for it. Most are layering agents onto the old process.

The top 10% of firms by usage now generate 8.3 times the output of typical firms, up from 2.6 times in January. OpenAI's own read is that access is not enough. Shared workflows, continuous learning, and governance are what widen the gap.

Across a three-year study of ten manufacturers, the skills that made new technology work emerged on the floor as workers and managers figured out how to fit the tools to the job. Training catalogs built on forecasts missed them.

Employees whose manager champions AI are about eight times as likely to say AI transformed how work gets done. Introducing AI on its own does not make a better workplace. A manager who leads it does.

Three years inside 11 IT services firms. Executives admit privately that AI requires real changes to pricing, staffing, and the business model, then drop those positions in the group. Reassuring stories replace hard decisions.

Researchers from Harvard, Wharton, Northwestern, and Columbia found generative AI tends to amplify the human bottlenecks inside innovation. Synthetic customers speed up testing and miss the irrational behavior that decides whether people buy.

87% of customers say a company using AI for service must offer a human. Asked what would change the mind of someone unwilling to use AI, the top answer was the ability to switch to a person.

82% of C-suite leaders are raising AI spend. 23% report widespread, sustained value, down from 32% earlier in the year. Accenture's own conclusion is that value comes from redesigning work, roles, and operating models.

Among occupation-specific work, 43.5% of what people ask AI to do belongs to a different occupation. Handoffs are disappearing. Job design either catches up or gets left behind.

US engagement is stuck at 31%. Where people use AI often, have a clear integration plan, and get active manager support, engagement climbs to 53%. Making AI available on its own does almost nothing.

55% of full-time employees use AI tools without approval or guidance. 68% want more leadership support to experiment. People are already redesigning their own work. The question is whether the organization joins them.

29% of knowledge workers admit to actively sabotaging their company's AI strategy, 44% among Gen Z. The top reason given is fear of losing their job. That is what a transformation looks like when nobody asked the people doing the work.

Companies redesigning end-to-end workflows were 24 points more likely to see measurable business impact than companies that only rolled out tools. Better tools on their own moved it about 5 points.

Treat AI as an IT rollout and the validation, error-catching, and coaching pile onto middle managers with no change to their delivery pressure and no new support. The fix is redesigning the manager's job.

The companies capturing the most business value are the same ones where employees enjoy work the most. Workflow redesign lifted job satisfaction 20 points alongside the business impact.

The bottleneck is usually old metrics, old assumptions, and workflows nobody questions because they have always been there. AI's real use is exposing where the process stopped making sense.

Culture, manager support, and talent practices explain twice as much of AI's impact as individual mindset and skill. Employees are ready to reinvent how they work. The metrics, incentives, and norms around them keep rewarding the old way.

HBR named the failure mode: the micro-productivity trap. Companies speed up individual tasks, never rethink the workflow or where value comes from, and wonder why nothing shows up at the P&L.

MIT researchers modeled AI adoption as a problem of task chains, not single tasks. Plug AI into an existing workflow and you get incremental gains. Regroup the tasks and remove the handoffs and the gains compound.

One fifth of companies capture three quarters of AI's economic value. They are twice as likely to have redesigned workflows around AI instead of adding tools, and their employees are twice as likely to trust what AI produces.

Half of employed Americans now use AI at work. In organizations that adopted it, about one in ten strongly agrees AI transformed how work gets done. Gallup's read: the workflows, roles, and processes never got redesigned.

A randomized experiment across 515 startups gave some founders one thing: examples of how other firms reorganized production around AI. Those firms found 44% more use cases, most in product development and strategy, and generated 1.9 times the revenue.

Across 51 real deployments, 77% of the hardest problems were organizational: change management, data quality, process redesign. Same technology, same use cases, very different outcomes.

Gartner expects more than half of enterprises to stop paying for copilots and smart advisors within two years and buy workflow results instead. Vendors that bolt AI onto legacy software face margin compression of up to 80%.

Oxford, Stanford, and BetterUp researchers argue automation shows early gains and augmentation wins later, because long-term success depends on how people feel about their work and whether top talent stays. 81% of senior leaders think they are augmenting. 40% of the front line suspects automation.

Roughly 15% of organizations use AI to fundamentally redesign how work is performed. The rest collect task-level wins that never turn into structural value.

Karim Lakhani and colleagues found the block is rarely model quality or data. It is where technical capability has to meet how the company is actually designed: pilots that never scale, process debt, tribal knowledge, the efficiency trap.

Organizations that intentionally design how people and AI work together are 2.5 times more likely to report better financial results and twice as likely to say they provide meaningful work. Only 6% say they lead here.

The most anxious employees use AI the most: 65% of their job is AI-assisted versus 42% for the least anxious. Fear drives compliance. It does not produce commitment, and it does not produce impact.

A UK bank took a 60 to 100 day, ten-handoff process down to a single day in four months by redesigning the workflow and the teams around it. Bain's warning: treat the people impact as downstream change management and AI never gets past micro-productivity.

In BCG's case work across hundreds of companies, about 10% of AI value comes from algorithms and 20% from the technology to run them. The other 70% comes from rethinking the people side. Managers at the leaders role-model AI at 88% versus 25% at laggards.

63% of employees are more likely to embrace AI when they understand how it is used and keep the ability to override it. In one bank, retraining analysts to challenge the system cut errors 40% and raised job satisfaction.

30% of organizations are redesigning key processes around AI. 37% use it at the surface with little or no change underneath. Only a quarter have moved even 40% of their pilots into production.

95% of organizations saw zero return on generative AI. The barrier was not infrastructure or talent. Most systems never learned the context of the work or fit into how it actually gets done.

Across four experiments, working with generative AI improved immediate output. When people went back to working alone, intrinsic motivation dropped and boredom rose. Design the collaboration for the person, or pay for it later.

Of 25 practices tested, redesigning workflows had the biggest effect on whether a company saw profit impact from generative AI. Only 21% had redesigned any.

C-suite leaders guessed 4% of employees use generative AI for at least 30% of their work. Employees reported three times that. McKinsey's conclusion: the biggest barrier is leadership.

Across eight product and service categories, describing a product as AI-powered lowered emotional trust and purchase intent. The effect was strongest for expensive and high-risk purchases. Describe the benefit. Leave the buzzword out.

64% of customers would rather companies did not use AI in customer service, and 53% would consider switching to a competitor over it. Their top worry is not being able to reach a person.

McKinsey tracked 300 public companies for five years. The top quartile on design grew revenue 32 percentage points faster and shareholder returns 56 points faster than their industry peers. More than 40% of companies still did not talk to end users during development.