Businesses that have an eye on increasing profits try to implement artificial intelligence in any number of ways, but in too many cases, their AI expectations are shattered.
RAND Corp. research reports that, by some estimates, more than 80% of AI projects fail—twice the failure rate of traditional IT projects. An S&P Global survey found that 42% of companies abandoned most of their AI initiatives in 2025, up from 17% the year before.
It is disappointment after disappointment, but the usual explanations for such dismal results—bad data, talent gaps, immature technology—miss the point. AI isn’t failing. Our ambition for what it should change is.
Start with the Experience You Want to Create
When companies launch AI initiatives, they almost always begin with the technology. They ask questions such as: What can this model do? Which processes can we automate? How do we implement it quickly?
This is exactly backwards, and the end-result is what I call the “back-office trap.” Businesses deploy AI to automate invoice processing, summarize internal reports, classify support tickets, or in other areas where failure is invisible, rather than in areas where success would be transformative.
These projects check the box, but deliver on narrow metrics and change absolutely nothing about how customers experience your business.
That’s why any change to the bottom line is minimal for so many businesses. Even though 88% of organizations now use AI in at least one business function, only 39% report any measurable impact on earnings at the enterprise level, according to McKinsey’s 2025 global AI survey.
But here’s a crucial finding from that survey: Organizations that reported significant financial returns are twice as likely to have redesigned workflows before they selected modeling techniques. They didn’t start with the technology. They started with the experience they wanted to create for customers.
Two Tales: One Cautionary, One Inspirational
Let’s look at two real-life—and very different—experiences with AI.
Klarna, the Swedish fintech, made headlines in 2024 when its AI assistant handled 2.3 million customer conversations in its first month. That was two-thirds of all customer service interactions, and the results appeared promising. Resolution times dropped from 11 minutes to under two minutes. The system operated in 35 languages across 23 markets, and the company projected $40 million in annual savings.
The AI was doing the equivalent work of 700 full-time agents, and it looked like the future of customer service.
As it turned out, it was not.
In mid-2025, Klarna’s CEO, Sebastian Siemiatkowski, admitted that the company had gone too far. The cost savings the AI assistant provided came at the expense of quality and at the expense of the customer experience. Everything had been technology-driven, not experience-driven.
Contrast Klarna’s initial approach with that of Delta Air Lines. In 2025, Delta unveiled “Delta Concierge,” an AI-powered personal assistant built into its mobile app. But instead of asking, “What can AI automate?” Delta asked, “How do we blend the digital and physical experience across the entire customer journey?”
The result isn’t a chatbot that handles service inquiries faster. It’s a system designed around moments that matter to travelers. Delta Concierge provides proactive passport expiration alerts before you book international travel. It provides real-time wayfinding to your gate. It tells you the shortest TSA PreCheck line or the nearest Sky Club, among other features.
The technology underneath Delta Concierge may be similar to what Klarna deployed, but the ambition is fundamentally different, as is the customer experience.
Regulated Industries: Designing within Constraints
Nowhere is the ambition gap more pronounced than in highly regulated industries. Sectors such as financial services, healthcare, and insurance often treat compliance as a reason to shrink their ambitions rather than as a design constraint to work within.
But leaders in these industries prove that regulation and transformation aren’t mutually exclusive.
JPMorgan Chase, operating in one of the most heavily regulated environments in business, has committed $18 billion annually to technology and deployed over 400 AI use cases.
Their approach is instructive. They started with high-value domains (credit decisions, fraud detection, wealth management) rather than low-risk back-office tasks. Their “Coach AI” tool helps wealth managers respond to client concerns during market volatility, contributing to a 20% increase in gross sales by identifying opportunities and enabling personalized strategies at scale.
The bank’s vision, articulated by Chief Analytics Officer Derek Waldron, is that “every client experience has an AI concierge.” That’s not a company hiding AI from regulators. That’s a company designing AI for the experiences that matter most, with clearly designed architecture and the appropriate governance built in.
Asking the Right Questions
AI’s transformative potential doesn’t lie in automating isolated tasks. It’s in rewiring the relational flows that connect your organization to customers, employees, and partners. When you view AI through this lens, the questions change entirely:
Instead of asking, “Which processes can we automate?” you ask, “Which relationships can we deepen?”
Instead of asking, “How do we implement this technology?” you ask, “What experience do we want to create?”
Instead of asking, “What tasks can AI perform?” you ask, “How should AI show up across the entire journey?”
At my company, Cortico-X, we took such an approach in developing what we call “Experience Architecture,” an operating model that fuses business intelligence, human intelligence, and artificial intelligence into one orchestrated system. It’s built on the idea that you start with the customer, not the technology.
Raising Your Sights
If you’re ready to escape the “back-office trap,” stop asking what AI can do and start asking what experience you want to create.
Before approving your next AI initiative, ask this: Are we designing an experience, or just automating a task?
If the answer is automation, you’ll join the 80% of AI projects that fail to deliver meaningful value. But if you start with the experience, designing AI as a relationship rather than deploying it as a technology, you’ll find that the returns aren’t just better. They compound.
The technology is ready for transformation. The question is whether your ambition matches it.
About Sujay Saha
Sujay Saha is the Founder and CEO of Cortico-X (cortico-x.com), an experience-led strategy and transformation business. His work centers on bringing a human-centric lens to solving business challenges and uncovering opportunities in today’s digital age. Previously, Saha served as a partner and leader of PwC’s Digital & Customer Strategy practice, where he worked with executive leaders across several industries, helping them translate growing customer focus and digital transformation into profitable growth. Saha has a bachelor’s degree in engineering from the Birla Institute of Technology, Mesra, and an MBA from the Cornell Johnson Graduate School of Management.
