By Rohit Kapoor, Chairman and CEO of EXL, a global data and AI company.
A decade ago, AI felt like a novelty. Something everyone talked about, but few understood. Today, it’s a practical tool shaping real-world decisions, in every industry, from insurance to banking to energy and utilities to healthcare to retail. The technology is completely transforming how companies operate. Yet many Australian businesses are struggling to capture this value, leaving leaders frustrated, teams stretched thin and investments at risk.
EXL research reveals that 60% of enterprise AI initiatives are still stuck in the pilot phase, unable to achieve meaningful scale or impact. Recent data from McKinsey, the U.S. Census Bureau, and S&P Global, likewise show that large corporations worldwide are struggling to move beyond AI experimentation to unlocking real-world value.
In fact, according to S&P Global, a whopping 42% of enterprise AI initiatives fail outright, which means nearly half of all companies that are investing in AI are unable to turn their investments into tangible returns. So, why is this happening? And more importantly, what can Australian businesses do to avoid becoming part of this statistic?
From Pilot to Impact
At EXL, we work with some of the world’s largest companies and the two biggest barriers that most often prevent orgs from delivering value at scale are siloed data and specialized talent gaps.
1. Siloed Data
AI models are only as good as the data that powers them. You may have heard of the ‘garbage in, garbage out’ adage. This phenomenon is especially true with AI. If the data feeding your models is fragmented, inconsistent or outdated, the insights you get back will be equally flawed.
Many organisations are still dealing with disconnected systems and incompatible platforms that make it difficult to bring data together across departments. Without a cohesive data estate and strong governance processes, it is nearly impossible to build AI solutions that can learn, adapt and operate effectively.
We’ve seen clients with incredible technical resources fail simply because their data lives in silos. The story always ends the same way. Endless pilots and prototypes that never make it into production.
2. Specialised Talent Gaps
AI success isn’t just about having access to the latest algorithms and tech gurus. True transformation only happens when AI is thoughtfully applied to complex, highly specialised workflows. Industries such as insurance, healthcare and banking are prime examples, where the right application of AI can dramatically improve accuracy, speed and decision-making.
This is where many companies stumble. They have brilliant technologists and the ability to buy cutting-edge technologies, but those teams often lack a deep understanding of the industries they’re trying to transform. For example, building an AI model for insurance claims processing requires not just machine learning skills, but also a deep understanding of claims workflows, regulatory compliance and customer experience.
When technologists and domain experts do not work hand in hand, companies end up with models that look impressive in testing but fail to deliver meaningful outcomes when they are applied in the real world.
From Pilots to Impact
While the industry average failure rate for enterprise AI initiatives sits at 42%, EXL has achieved a 94% success rate in embedding AI directly into core business workflows. That’s not by chance. We’ve learned that successful AI adoption depends on getting three things right: data, domain and AI.
These are what we call the three pillars of AI success. When data is well-governed and accessible, when there’s deep domain expertise to guide how AI is applied and when advanced AI capabilities are integrated thoughtfully, companies can move beyond pilots to achieve meaningful, scalable impact.
This formula consistently moves our clients beyond endless pilots and into full-scale implementations that deliver measurable business outcomes. It’s also the reason EXL has delivered 20 consecutive quarters of earnings growth, including five straight quarters of double-digit expansion.
One example of how our unique approach is playing out in the real world is a project we’ve been developing with a leading Australian insurer. We have deployed our Smart Data Signals solution to continuously monitor all claims activity and provide real-time alerts whenever our application spots an anomaly or inconsistency. That allows everyone involved in the underwriting and claims process to intervene quickly and avoid the types of errors, inefficiencies and inconsistencies that caused costly leaks in the claims cycle.
The Power of Partnership
This AI success does not happen in isolation. EXL partners closely with the world’s leading technology providers, such as Nvidia, Microsoft, AWS, Google, and Databricks, to help clients access powerful tools that can enable adoption at scale. We play a key role with these much larger players because we are driving expanded usage of their technologies and their cloud platforms. As our customers use these solutions more widely, we end up helping the whole ecosystem expand.
Our work with technology partners enables us to deploy these cutting-edge tools directly in our customers’ environments, ensuring best practices. We act as an enabler to translate the technical capabilities into domain solutions. This is where the domain, data and AI come together. Put simply, technology companies bring the tools and we bring the expertise to make them work in the real world. That’s the bridge Australian enterprises must cross to turn AI ambition into tangible business outcomes.
The Big Australian Opportunity
Australia has a long history of innovation that keeps pace with the world, but right now, there’s a growing AI adoption divide. Companies that move quickly to embed AI into their workflows will gain a significant competitive advantage. Those that stay stuck in pilot mode risk falling seriously behind global competitors who are already using AI to reimagine their business workflows to improve efficiency and reduce costs.
