Throughout 2025, I spoke with countless business leaders about their AI strategies, seeking to understand what was working for them and what was holding them back. Over the year, I’ve noticed three trends that have continued to emerge, across companies and industries, determining which companies are succeeding with AI and which are struggling. I now bring these trends together, offering lessons from the front lines of AI transformation.
First, the use of AI for back-end tasks is booming, showing that it’s often the boring tasks that can really shake things up. The second trend is not about technology, but rather about people: how companies approach their employees is critical to how AI adoption unfolds. However, perhaps the most telling trend concerns initial strategy and motivation. Businesses fail when they rely on AI and succeed when they tackle the problem they are trying to solve.
Of course, there’s much more to it, from data management to security and governance. But these aspects shape AI efforts, for better or worse.
Avoiding AI for AI’s sake
Erik Brown, head of AI and emerging technologies at consultancy West Monroe, said Fortune Earlier this year, he saw many companies struggle with “AI fatigue” after becoming frustrated with unsuccessful AI proofs of concept. The common theme among those who have found themselves in this position, he said, is that they explored the wrong use case or misunderstood how AI might (or might not) be relevant to the task. Specifically, they started with the idea that they wanted to pursue AI, rather than the problem they wanted to solve.
For example, he said one client brought together its best data scientists to form a new “innovation group” to figure out how to deploy AI, only to end up wasting tons of resources on ideas that were cool but didn’t solve any real problems for the business. After his team suggested the company take a step back and ask business units to identify key challenges, the consultants quickly discovered an area where AI could really help, proved it by working hand-in-hand with the business unit, and deployed the solution.
“I think it’s so easy with any new technology, especially one that has AI attention, to just drive the technology first,” Brown said, echoing an observation I’ve heard repeatedly throughout the year, including from business leaders and other consultants helping companies navigate AI transformation.
BigRentz is a company that demonstrates the other side of the coin. The construction equipment rental company remained hyper-focused on the problem it was trying to solve and ended up reinventing its entire business using AI. CEO Scott Cannon said Fortune they “didn’t set out to build our business around AI. It just turned out to be the best tool for the job.” Additionally, BigRentz used only old-school machine learning, showing that even in the era of generative AI buzz, previous AI techniques still have value and why it’s important to find the right solution for the right problem.
Honeywell is another company that launched each project with a clear strategy for what it wanted to accomplish, having created a meticulous framework to guide the development and deployment of its AI. It has paid off: every function and strategic business unit in the company now uses generative AI, and the company has 24 generative AI initiatives in production and 12 more in progress, up from 16 a year ago.
“What are the use cases? And can I measure and track them?” Technical Director Suresh Venkatarayalu said Fortunedescribing how the company starts with value-add when considering any potential AI effort.
Boredom gets results
The idea of avoiding “shiny object syndrome” is good advice, especially as the AI hype quickly shifts from chatbots to agents and then whatever comes next. Another reason not to follow the latest hype: Many organizations are discovering that it’s the boring, back-end uses of AI that really make the difference.
Law firm Troutman Pepper Locke is using AI in a variety of ways, including creating its own AI chatbot assistant that all employees can use. But the director of innovation, William Gaus said Fortune The company currently believes that AI is most useful for back-end administrative tasks, which it says are also a great place to start because they are low risk.
For example, when the firm was finalizing its recent merger, its team created an agent capability to revamp the biographies of the 1,600 new attorneys, which needed to be updated to include the new firm’s information and match its existing writing style. Gaus described how this made the process significantly more efficient compared to the last time they undertook this task, which required six months of manual labor. In total, the company saved $200,000 in time spent, he said.
The same thing is happening in the medical field. Efforts to create a reliable healthcare companion chatbot, for example, have made little material progress. But AI tools are deployed in the back ends of the healthcare system. Doctors use LLMs to record and transcribe conversations between themselves and patients to generate medical documents, allowing them to be more engaged with the patient and reducing time spent on paperwork outside of their work hours. They also use LLMs to quickly create synopses of complex medical records and more easily query medical databases.
“The things that we’re taking off the clinician’s plate, which are more administrative, I think are some of the areas where we see AI evolving very quickly,” Wiljeana Glover, a researcher focused on healthcare innovation and improvement at the Kerry Murphy Healey Center for Health Innovation and Entrepreneurship at Babson College, said Fortune.
Keeping people first
For all the talk about use cases and business strategy, we must not lose sight that people are at the center of AI transformation. It’s unclear whether AI is leading companies to lay off employees. still not clear…and if it doesn’t currently, that doesn’t mean it won’t change in the future. Yet today, AI is already having a huge impact on people in their work, in the way they work. hiring And qualified the tasks and expectations assigned to them. How companies navigate current changes and concerns about the future directly impacts how employees approach AI transformation.
Perhaps more than any other term, the leaders I spoke with this year talked about “change management,” referring to how an organization moves from its current state to a desired new form with maximum successful adoption and minimum disruption.
Honeywell’s other head of AI, senior vice president and chief digital technology officer Sheila Jordan, warned: “You can’t underestimate it. » Accenture Lan Guan, Director of AI, suggested that a company can create all kinds of amazing AI tools that solve business problems, but it’s just as important to make sure your employees are ready and open to using them. Others talked about the need to bridge the gap between overzealous AI supporters (who might chase AI for AI’s sake) and AI skeptics.
A key part of this is for business leaders to keep their promises and expectations of what AI can deliver in check. Some software developers and engineers – the first cohort to see their work truly disrupted, thanks to the proliferation of AI coding tools – say they are frustrated with how many leaders are overselling AI and inflating what it can do. Others I felt overwhelmed by unrealistic expectations to produce more code more quickly or to use specific tools, disappointed by the mandates set by managers who do not understand the daily life of their work and who prioritize productivity above all. When these changes are announced by tech leads with hands-on experience, or even by the developers themselves, it often leads to more positive sentiment and better results.
Even though all the productivity in the world is possible through AI, some executives are wary of what too much outsourcing of work, especially entry-level work, will mean for the workforce in the near future. For example, Ryan Anderson, co-founder and CEO of Filevine, a company that creates AI tools for the legal industry, said he worries about young lawyers using AI co-pilots being able to develop their creativity and ability to gather information on their own.
“An over-reliance on AI,” he said, “could be just as problematic as the exciting opportunities it brings.” Finding the right balance should be one of the key agenda items as businesses advance AI in 2026.