AI scaling issues are hiding in plain sight. Most companies have approved budgets and started pilots, but few are seeing meaningful results. Using generative AI is great, but most users and businesses are only scratching the surface when it comes to the potential it can unlock.
McKinsey’s 2025 State of AI Study confirms this. Today, 88% of organizations are using AI in at least one business function, but only 6% are recognized as “AI High Performers” who are creating meaningful impact. There is a gap between successful pilots and scaling AI across the organization, and most efforts fail.
This is not a technology issue. It is commercial and institutional.
Notes on perspective
For the past 20 years, I have worked in venture capital and private equity as a commercial and business leader. I invest in and evaluate dozens of companies, from early-stage AI natives to late-stage traditional technologies, and advise portfolio companies on accelerating revenue growth. This gives you perspective from both sides. We see AI vendors trying to acquire enterprise customers and businesses trying to make AI work at scale.
Why most AI pilots can’t scale
When looking at stalled AI efforts, the same problems keep coming up over and over again.
The first is a lack of business transparency. Too many initiatives start with the question, “What can we do with AI?” Instead of “What business problem are we trying to solve?” It sounds obvious, but it’s very common. In my experience, more than half of generative AI initiatives fail to get past proof of concept. Teams create impressive demos that demonstrate technical capabilities, but lack clear connections to revenue, costs, and customer outcomes. As budget decisions approach, these initiatives struggle to justify continued investment.
Second, the workflow remains unchanged. Rather than rethinking how work gets done, companies are bolting AI into existing processes. The real value comes from redesigning the work that AI enables. Most companies can’t get there.
Third, there is a lack of investment in change management and skills. Companies spend money on technology, but not on the people side of transformation. For every $1 spent on technology, we may need to spend $3 on helping people adapt. McKinsey identifies similar ratios as critical to successful AI transformation. But most companies ignore it. They assume that people will adopt new tools because they are better. This rarely happens. Resistance, confusion, and old habits slow retention.
Meanwhile, most employees lack training on how to use AI effectively. The technology is ready, but the organization is not. In fact, one of the most prominent patterns I see is that individuals are silently using AI tools on their own because it takes too long to get corporate buy-in.
If your employees are using AI to solve problems faster than they can be approved in the procurement process, it’s not a technology gap. It’s an organizational thing.
What makes great enterprise hires different?
Companies that are successfully scaling AI share a different set of behaviors.
They don’t pursue use cases that have a bigger impact. Rather than spreading their investments across dozens of experiments, they focus on a small number of applications with clear business value, where the potential impact is measurable. Depth over width.
They don’t just automate the work, they redesign it. Instead of thinking, “Where can we add AI?” they ask, “How should this function in an AI-native world?” The productivity gains from redesign are many times greater than those gained from automation alone.
Klarna explains both the opportunities and the nuances. Rather than incorporate AI into its customer service operations, the company redesigned its entire model. The company’s AI assistants now handle two-thirds of all conversations, doing the equivalent work of more than 850 agents and saving the company $60 million. But in mid-2025, the CEO admitted that he had placed an “over-emphasis” on AI at the expense of service quality and began rehiring human agents. Redesigning work around AI will yield transformative results, but the human element will remain essential, at least for now.
Measure business outcomes rather than traditional adoption metrics. Tracking the number of people using AI is a good sign, but at the end of the day it comes down to measuring impact: did customer satisfaction improve, did processing times improve, did error rates decrease? This distinction is important because deployment without impact is just a cost.
They treat talent and training as a major investment rather than an afterthought. Change management is the real cost of AI transformation, and companies that get it right scale faster.
What AI technology companies are doing to win
The smartest AI vendors recognize that technical capabilities alone cannot close deals or drive adoption. Enterprise buyers get burned by software that doesn’t scale and ends up sitting on the shelf. Winning in this market requires a different approach.
Great vendors are shifting from selling features to solving specific, persistent problems that keep customers up at night and drive compelling events to act now rather than later. They lead by results, not ability. They can explain exactly how their products reduce costs, increase revenue, and reduce risk. Those who get real traction do well
or. They root their solutions in industry-specific pain points that customers immediately recognize. When vendors can talk about specific challenges in an area, rather than offering generic horizontal tools, the conversation changes from “Why AI?” “How quickly can we implement it?” That specificity creates value not only for companies but also for customers. This creates a powerful flywheel.
I see this firsthand through the companies we invest in. Two examples illustrate the pattern.
The first is an AI platform that saves pharmaceutical companies up to 18 months of drug discovery time by automating clinical trials. The company has built the space’s first true AI-native platform to automate key clinical trial processes and patient recruitment. With major pharmaceutical companies on board as customers, the platform achieved 21x ARR growth in one year. This works because the founders understood a concrete, high-value problem: that clinical trials have historically been done manually and are extremely time-consuming. They created a purpose-built solution rather than a horizontal tool looking for use cases.
The second is an AI platform that automates clinical management for healthcare providers, including taking notes, handling patient calls, managing emails, and processing insurance claims. Clinicians who use it save more than two hours per day, and clinics report a 30% reduction in administrative costs. Clinical compliance, data security, and integration with existing medical record systems were built into the product from the beginning, rather than as an afterthought. This has given healthcare organizations the confidence to incorporate it into their daily operations.
None of these companies build general-purpose AI tools and then start looking for problems to solve. They started with a specific industry, a specific pain point, and built a purpose-built solution around that. In most cases, the founders have experience in the industry. They themselves feel the pain. Once you sit down with the problem, it becomes much easier to build the right solution. That’s a winning pattern.
Particularly in highly regulated and complex fields such as healthcare and financial services, barriers to entry are high and tolerance for failure is low. It is that complexity that creates opportunity. Vendors that do the hard work of building for these environments create defensible positions that horizontal AI tools simply cannot replicate. It’s also where I focus most of my investment. Because moats become real and value increases when companies solve painful industry-specific problems in regulated markets.
The best vendors not only solve the right problem, but also get the commercial model right. They build governance and compliance in from day one rather than adding it later. They design workflow integrations so that their products fit into people’s existing ways of working. They are moving to usage and outcome-based pricing that matches cost and value. And we invest in our customers’ success because we lose trust when something works in the demo but doesn’t work in the real world.
The difference between AI experimentation and AI at scale will be the difference between gaining or losing competitive advantage in the coming years. And we’re in a very strong period for AI businesses. The companies that close it aren’t the ones with the most advanced models or the biggest budgets. They will treat scaling AI as a business transformation rather than a technology project.
The technology is ready. The question is whether organizations are ready to change the way they work. If they don’t take action, they will not only miss out on the financial turnaround, but also be left behind by competition and market changes.
Mitul Ruparelia is General Partner of Arãya Ventures Global Fund
Main image source: iStockPhoto.com and Dakuku
