AEGIS London Achieves 98% Accuracy with GenAI-Enabled Endorsement Processing Facilitated by Fisent BizAI

How a leading UK property and casualty insurer, AEGIS London, with an annual turnover of more than $1 billion, achieved transformative efficiency and underwriting insight by leveraging Fisent BizAI, the industry-leading Applied GenAI Process Automation solution.

The Challenges

Across enterprises, manual, human-dependent tasks are creating significant inefficiencies. These tasks, often involving diverse information, interrupt workflows and divert skilled personnel from core responsibilities and more valuable work. This is a universal problem for information-reliant organizations, where data variations make manual processing cumbersome and prone to error. AI holds a compelling solution to this challenge.

According to Roger Misra, Change Leader at AEGIS London, the rapid advancement of AI prompted the company to seek new capabilities. “We were getting lots of requests from our leadership team to try to add new AI capabilities within our key underlying processes,” he stated.

Like many in the property and casualty insurance industry, AEGIS London faced a critical bottleneck in its policy endorsement process. Processing endorsements, which are mid-year policy changes, previously required significant manual effort from brokers and underwriters. They had to interpret information, determine additional premiums, secure agreement on changes, and manually capture data for downstream systems. The core challenge was the sheer volume, tens of thousands of annual inquiries, which came from insureds and insurance agents, arriving in various formats with inconsistent details, making the workflow laborious and inefficient.

Experienced underwriters must manually analyze inbound inquiries to extract critical information. They then compare these details against existing policies to determine the necessary adjustments, summarizing the required policy amendments. This can involve adding, deleting, modifying, or excluding coverage to address the unique circumstances surrounding each policyholder’s change. This manual workflow was time-consuming, with each endorsement taking an average of five minutes.

AEGIS London correctly expected a GenAI solution should be able to automate a significant portion of endorsement processing. It needed a partner that could not only deliver a solution but also integrate seamlessly with its existing Pega platform and provide a clear path for future efficiency in a highly regulated environment. To address this challenge, the company took on a complex vendor selection process. The company’s comprehensive RFP revealed a wide range of options, from legacy market stalwarts with rigid, pre-trained models to consultants offering generic, platform-based solutions.

Finding Fisent BizAI

After a thorough evaluation, AEGIS London selected Fisent as its partner, viewing the company as the ideal fit. While AEGIS London considered a competitive solution focused solely on the London market, the company believed Fisent would be the better long-term choice due to its demonstrated ability to solve a wide range of complex problems. Roger Misra explained the choice, stating that Fisent is “a highly agile operation with knowledgeable people and proven ability to navigate the corporate governance we require.” The partnership began with an accelerator sprint, a hackathon-like workshop where key AEGIS London stakeholders, including underwriters, developers, and analysts, collaborated to prove the concept.

The team ultimately chose to tackle the policy adjustment endorsement process as the first at-scale manual process to automate. In only six weeks, Fisent BizAI was configured to automate rule application, understand the full rule set underwriters use to categorize endorsements (e.g., early, material, read-only), infer context, and capture the key data points from endorsement documents. The implementation was highly iterative, beginning with just six rules and ultimately expanding to a robust 55 rules. A critical element of the project’s success was a swift feedback loop, which allowed the administrative team to provide direct input on performance, leading to continuous refinement and improvement, and improving accuracy.

Business Outcomes

The implementation of Fisent BizAI yielded significant benefits for AEGIS London, transforming its endorsement process and setting the stage for future innovation.

• Increased Accuracy: Fisent BizAI’s accuracy in categorizing endorsements and extracting data fields improved from around 70% to 98% after just three iterative cycles of refinement.
• Dramatic Time Savings: The time to process a single endorsement was reduced from an average of five minutes to an average of about two minutes. An upcoming API integration is expected to further reduce the average processing time to between 15 and 20 seconds per endorsement. This integration is projected to automate more than 90% of all endorsements.
• Enhanced Underwriter Efficiency: AEGIS London is now able to achieve a quicker and more consistent turnaround time, allowing underwriters to focus on more high-value tasks.
• Improved Transparency: Fisent BizAI transforms unstructured or semi-structured data from documents into a structured format, which provides a better view for reporting and improves market knowledge

Uncovering an Unknown Risk

A critical, unexpected outcome of implementing Fisent BizAI was the discovery that a quarter of all endorsements were going unprocessed. These documents were simply sitting untouched in underwriter mailboxes, creating a lack of transparency and a substantial risk to business continuity.

While the immediate benefit was a reduction in exposure through automated routing, the long-term impact on AEGIS London’s risk posture is even greater. Fisent BizAI is now providing a rationale for all of its analyses. This new, data-driven insight helps the company better understand and manage its exposures, creating a more robust and transparent process than was previously possible. 

More Human-Dependent Tasks to Automate

Fisent’s agentic solution can be applied to a multitude of at-scale manual tasks. Most organizations typically find two to three dozen of these automation gaps within their operations.

For AEGIS London, the success of the policy adjustment endorsement automation effort has led to the implementation of Fisent BizAI throughout its operation, including document ingestion for new insurance quotes and the creation of risk summaries. AEGIS London views its relationship with Fisent as a long-term partnership, with the goal of implementing three to four new use cases per year.

According to Roger Misra, the return on investment for Fisent BizAI was evident, as it not only improved a single process but also provided a robust solution for ongoing digital transformation. He further noted that there were numerous potential applications for Fisent BizAI cases.

Key Takeaways

AEGIS London provides a clear example of how a major insurer, through its adoption of Fisent’s BizAI solution, transformed a cumbersome manual process into a highly efficient and accurate automated workflow. Key takeaways include:

• Fisent BizAI achieved a 98% accuracy rate in data extraction and categorization, ensuring a high level of reliability.
• Endorsement processing time was dramatically reduced, from an average of 5 minutes to about two minutes, with the aim of driving that number down to 15-20 seconds per endorsement.
• Underwriters are freed from repetitive tasks, allowing them to focus on more high-value work.
• The Fisent BizAI implementation identified that 25% of all endorsements were going unprocessed, eliminating a significant risk.

Ultimately, the success of Fisent BizAI proves the power of Applied GenAI Process Automation to drive transformative business outcomes and provides AEGIS London with a scalable solution for future digital transformation.