How quickly have enterprises across industries embraced AI? The answer depends on your view of its origins. Some people believe AI dates back to Alan Turing's early theories of machine intelligence. Others see the Dartmouth Summer Research Project on AI in 1956 as its birth.
What is unarguable, however, is that AI has moved at breakneck speed since the launch of ChatGPT three years ago. A recent McKinsey survey revealed that 88% of organizations now use AI regularly. Yet there is a downside. Only one-third of respondents have scaled AI enterprise-wide, and fewer than half can tie any significant earnings before interest and taxes (EBIT) to their efforts.
What will it take for insurers to see real ROI from their AI investments? And how can carriers prepare for an agentic AI future? A look at past tech trends can show us quite a bit about what is to come.
Why Every Tech Revolution Stalls Before It Scales
All significant technological breakthroughs of the past 20 to 30 years have followed a distinct progression. First comes the hype, where inflated expectations generate massive excitement and speculation. Next comes normalization, where practical use cases emerge. Lastly, there is substantive change, where technology becomes embedded in core operations, driving genuine business value.
Within each cycle, however, key barriers exist that block progress once the initial hype fades. In the late 1990s, high-speed connectivity was going to reshape business overnight, but ROI did not come until broadband access expanded. Apple's first iPhone launched in 2007 but did not achieve widespread use until its app store and developer ecosystem matured. Cloud computing brought huge promise in the mid-2000s but did not mature fully until operating models shifted toward DevOps, APIs and microservices.
Even insurtech itself went through a metamorphosis. Early disruptors that set out to overturn insurance incumbents about a decade ago have either been acquired or failed to prosper. Those that focused on partnering with incumbents to modernize core parts of the insurance value chain, however, have grown stronger and manifested an industry that drives innovation in insurance forward.
What's Holding AI Back?
There clearly is no shortage of excitement or action when it comes to implementing AI in insurance. Seventy-six percent of insurers already use generative AI in at least one core business function, making insurance the second-fastest industry to AI adoption, trailing only tech, according to BCG research.
Yet true transformation value remains rather elusive. Despite the breakneck speed of adoption, the results of AI pilots are uneven, and progress is patchy. Only 6% of respondents to McKinsey's survey are classified as high performers. That means they set targets beyond efficiency, embedding AI in existing operating models and new products to capture more revenue and expand cross-selling opportunities.
These high performers have already moved beyond generative AI and toward agentic AI, which involves using AI-powered agents to plan and execute multi-step work. Sixty-two percent of McKinsey respondents are already experimenting with agentic AI, and 23% are scaling agents somewhere in the enterprise. Here again, however, penetration is narrow, with agentic deployments living in one or two functions.
So, what's holding insurers back from ROI with generative and agentic AI? One common thread is deployments that are saddled by the limitations of legacy and mainframe systems. Put simply, insurers cannot reach an AI-enabled future if their workflows are dependent upon decades-old technology that is inflexible and impossible to integrate.
Breaking Through the Barrier
To deliver meaningful returns from their AI bets, carriers must rethink the foundations supporting their technology ecosystems. Many of today's challenges stem from years of product-centric digital transformation that delivered incremental wins but left insurers with a patchwork of disconnected systems. These solutions were never designed for the real-time data or observability that AI and agentic AI require.
A platform-centric approach offers a practical path forward. Unlike point solutions that focus on single steps in the value chain, modern platforms provide the connective tissue that allows data, workflows, models and controls to operate together. Platforms succeed because they allow insurers to redesign workflows holistically; something point solutions were never built to support.
The most effective platforms in the coming three to five years will share several attributes.
- Flexible, API-driven architectures that allow carriers to integrate new technologies into existing systems without a costly rip-and-replace.
- A broad ecosystem of partners and integrations, giving carriers access to specialized tools, such as distribution and fraud detection, without adding further fragmentation.
- AI-enablement at the foundation, meaning agents and models can operate safely and consistently across underwriting, claims, servicing, finance and product development.
- Built-in governance and human-in-the-loop patterns that build transparency and trust as insurers ask AI to take on more complex work.
- Outcome dashboards that provide clear metrics, including cycle time, loss adjustment expense, bind ratio, Net Promoter Score, and revenue uplift per use case.
With a platform-centric mindset, insurers can begin building momentum with their existing AI investments and gradually move toward an agentic future. A smart approach is to embed AI agents into crucial workflows where it can consolidate data, reason about actions, and hand off smoothly to humans, such as quote triage in commercial lines or claims final notice of loss (FNOL) to settlement recommendations. Start small, measure the impact, and then implement AI agents into other projects to make adoption stick.
Navigating the Road Ahead
From the dot-com boom to the rise of AI, history shows us reinvention in insurance happens over time, not overnight. AI is moving faster than any wave before it, but the core principles for insurtech companies and carriers alike remain the same: build and implement platforms that solve real business problems and engender trust with customers along the way. This type of measured approach will propel the next generation of AI-powered insurtech platforms and help accelerate the AI adoption cycle.
