Like virtually every other industry, insurers worldwide are looking at incorporating generative artificial intelligence (GenAI) into their business. In fact, EY found that 42% of insurers are already investing in GenAI and 57% have plans to do so soon. However, many insurance companies are likely to run into a major obstacle when it comes to generative AI integration: their legacy systems.
Legacy systems play a critical role in the current operations of many insurers, but they typically lack the flexibility needed to integrate with modern technologies like GenAI. This type of technical debt can be a major factor when digital transformation strategies fail.
Given that leveraging GenAI and other forms of AI effectively could deliver up to $1.1 trillion of annual value in the insurance industry, navigating the complexities of integrating GenAI with legacy systems is crucial to ensure that GenAI can be part of your company’s transformation journey. This blog will explore the challenges and solutions of these types of integrations, with a focus on minimizing disruptions and ensuring seamless operations.
Challenges of Legacy System and Generative AI Integration
As insurers grapple with how to use GenAI — and how it could benefit their customers — their legacy systems might prevent them from getting started. There are three significant challenges to GenAI adoption that carriers need to consider:
- Technical Compatibility
Fifty-seven percent of insurer decision-makers cite legacy IT system integration as a significant hurdle when exploring GenAI. One of the primary challenges in integrating generative AI with legacy systems is technical compatibility. Legacy systems often use outdated data formats, protocols, and infrastructure that are not readily compatible with modern AI technologies. Fragmented data infrastructures, siloed systems, and limited analytical capabilities can all lead to significant delays in implementation and integration failures. - Data Quality and Management
Legacy systems often have incomplete or outdated data, data silos, and inconsistent data formats, all from various sources. This makes establishing an accurate data repository to power a GenAI platform incredibly difficult. Not only can incomplete or inaccurate data hinder the platform’s ability to provide applicable insights, but it could also harm an insurer’s relationship with its customers.Recently, a national auto insurance company had to navigate data complexity when it started using GenAI. It possessed a lot of data but faced problems processing and analyzing it to derive insights. The insurer struggled to provide personalized recommendations because of the vast amount of data it had. As a result, the company turned to a partner that used both GenAI and machine learning so it could process the data using a more robust platform, saving time on developing AI models. - Organizational Resistance
In a survey late last year, seventy-one percent of U.S. workers who are familiar with AI reported feeling concerned about the technology’s impact. Almost half of respondents said they were more worried about AI now than a year before, while 41 percent said it’s evolving too rapidly. Employees and management may resist AI implementation due to a lack of understanding, fear of job loss, or disruption of established workflows. This organizational resistance can slow down or completely derail AI integration efforts.
With these challenges in mind, it may seem like insurers are facing obstacle after obstacle before they can adopt GenAI. Luckily, there are a few solutions insurers can use to combat each of the above challenges.
4 Ways to Succeed at Integrating Legacy Systems and GenAI
By strategically planning and adopting the right solutions, insurance companies can successfully integrate GenAI into their legacy systems, enhancing their operations, improving customer experiences, and staying competitive as the market evolves.
Here are four ways to ensure a successful integration of GenAI with legacy systems.
-
Implement a phased approach to integration
Using a phased approach to GenAI integration can help manage risks and disruptions. A phased approach involves starting with pilot projects, gradually scaling up, and continuously gathering feedback to refine the integration process. This method allows for addressing issues incrementally and minimizing unexpected challenges or gaps, ensuring a smoother transition.
Create guidelines around the initial use case that will be used to pilot the GenAI implementation. These guidelines should include security and risk evaluations so it is easy to understand where data is vulnerable. Then, implement GenAI into the pilot program to determine its value for the organization. This phased approach allows any roadblocks to be addressed before full implementation.
- Explore middleware solutions to bridge the gap between legacy systems and GenAIMiddleware and application programming interfaces (APIs) are intermediaries that facilitate communication and data exchange between incompatible systems. Middleware translates and manages communication between disparate systems, ensuring that GenAI solutions can seamlessly access and process data from legacy systems.APIs, on the other hand, allow insurers to expose specific functionalities of their legacy systems, enabling GenAI applications to interact with existing data and processes without requiring extensive rewrites or overhauls. One common approach to enabling legacy systems to connect with the modern world is by using the “Strangler Fig” This modular approach helps preserve the stability of core systems while introducing advanced AI capabilities incrementally.
- Improve data quality and consistency to support AI initiatives.Data modernization strategies, such as data cleansing, consolidation, and standardization, ensure that the data fed into AI models is accurate, consistent, and up to date.
One insurance firm in the Nordics needed to update its claims management process, for example. Its current workflow required team members to manually process each claim request, and even worse, its data was unstructured and came from various sources that required agents to examine and categorize the information closely. With so much data in so many different places, customers suffered from not receiving the answers they needed due to inefficiency and the inability to automate, thanks to the company’s legacy systems.
First, the insurer leveraged machine learning to help review the structured and unstructured data to improve the quality of the data available. Once the data was structured, it was transferred to the insurer’s legacy system, allowing the company to process that data in real time using AI. This helped the insurer reduce manual tasks by up to 70 percent, giving agents more time with their customers. These types of use cases will only be strengthened as insurers develop more and more capabilities fueled by GenAI.
- Address resistance with effective change management and training programs.Effective communication strategies, stakeholder engagement, and continuous training can help employees understand and embrace AI technologies. These initiatives ensure employees are aware of GenAI’s integration into their work systems and equipped with the necessary skills to enhance their day-to-day. Utilizing models such as Prosci’s ADKAR ensures that people think about the steps someone needs to really adopt change. By providing comprehensive training, insurers can address potential knowledge gaps, reduce resistance to change, and build employee confidence as they use AI-driven tools.
One of the largest insurance companies in the world used change management strategies to help ensure GenAI adoption at its organization. It involved extensive employee training programs, leadership communications that outlined the strategic benefits of GenAI, and testing programs that reached over 4,000 employees in its initial deployment. These strategies helped the insurer’s employees understand and learn how to engage with GenAI tools effectively.
Integrating GenAI with legacy systems presents both challenges and opportunities for transformation at P&C insurers. Technical compatibility, data quality, and organizational resistance are significant barriers, but insurers can overcome them with the right strategies. By navigating these challenges — with the help of a phased approach, middleware, data modernization, and change management — insurers can unlock the full potential of AI technologies and drive significant advancements in their organization.
Interested in learning more about using data modernization and tools like AI in insurance? Read our whitepaper, “Driving Business Value With Insurance Data Analytics.”