Suncorp Accelerates Claims Processing: How AI Agents Cut Time and Boost Accuracy by 99%

Suncorp, one of Australia's largest insurance giants, has fundamentally transformed its approach to claims settlement. By implementing five AI agents to handle various "sub-processes," the company has achieved not only a significant acceleration of work but also an increase in decision accuracy to 99%, even in complex tasks like coverage verification and incident classification.

Insurance claims processing is a tangled mess of routine and human error: receiving claims, classifying them, verifying coverage, assessing damages, coordinating with contractors, and finally, making payouts. Each of these stages is a potential point of failure, leading to delays, customer dissatisfaction, and direct losses. Today, this tangle can be unraveled by handing routine tasks over to AI agents, leaving only truly complex cases to human intervention.

The Problem: Where Suncorp Lost Time and Money

Before the implementation of AI agents, Suncorp's insurance claims processing was typical for a large insurance company: complex, multi-stage, and extremely resource-intensive. Thousands of claims passed through many hands and systems daily, with each stage requiring manual processing, verification, and decision-making.

The main problems were:

• Slow processing speed. Every step, from initial claim intake to final payout, took time. This led to significant delays, customer frustration, and increased operational costs.
• High risk of errors. Human error is inevitable. Mistakes in classification, document verification, or payout calculations could lead to financial losses, legal risks, and reputational damage.
• Inefficient resource utilization. Highly skilled professionals spent a significant portion of their working hours on routine, repetitive tasks that did not require deep expertise but consumed valuable time.
• Difficulty in scaling. During peak loads, for example, after major natural disasters, the system struggled to cope with the volume of incoming claims, leading to even greater delays and customer dissatisfaction.

Suncorp, as a company that has long used data and AI in its operations, recognized the need for a radical change in approach, making the process more "agent-driven," where specialized AI agents could take on routine tasks.

The Path to AI Agents: From Automation to Autonomy

Suncorp already used various automation and AI tools, but they were often separate scripts or models performing narrow functions. They could not autonomously coordinate actions, make decisions, or adapt to changes. What was needed was not just a tool, but a system capable of acting autonomously, mimicking human logic but with much greater speed and accuracy.

This is why the company adopted the concept of AI agents – specialized software entities capable of performing specific tasks, interacting with other systems, and making decisions based on defined rules and data. This would allow not just the automation of individual steps, but the creation of an entire "orchestration" of the claims settlement process.

How AI Agents Were Designed for Suncorp

The Suncorp team designed an architecture of five AI agents, each specializing in a specific stage of claims settlement. Their functionality:

• First Notification of Loss (FNOL) Agents. Two of the five agents focus on the very beginning of the process. One of them is presumably voice-enabled, taking calls and independently collecting and verifying initial information. The second agent classifies the incident, determines its type, and routes it to the correct department or to the next agent for further processing.
• Coverage Verification Agent. This agent is responsible for the critical step of determining whether a claim falls under the terms of the customer's insurance policy. It analyzes the policy text, compares it with incident details, and makes a judgment, significantly reducing the time for manual verification.
• Contractor Dispatch Agent. Once a claim is approved, this agent automatically dispatches requests to relevant contractors, such as repair teams or emergency services, to the incident location.
• Payout Calculation Agent. The final agent in the chain automatically determines the amount of compensation based on collected data, damage assessment, and policy terms, preparing the final calculations for the customer.

This entire system is orchestrated using BPMN (Business Process Model and Notation) workflows, where each task can be performed by one or more AI agents, ensuring flexibility and transparency.

Implementation and Control: Reliability and Security

The implementation of AI agents took place in stages, starting with the least risky and most labor-intensive processes. Kranti Nekkalapudi, Executive General Manager of AI at Suncorp, emphasized the importance of monitoring and risk management. For this purpose, a centralized observability platform based on Databricks was developed, collecting real-time data from all agents and workflows. This allows for:

• Performance tracking. Leadership can see the big picture, business users can see details of specific decisions, and engineers can debug the system.
• Ensuring security. The platform monitors agent operations, identifying anomalies or potential attacks.
• Guaranteeing accuracy. Suncorp developed a library of 100 different controls for various types of AI applications, achieving 99% accuracy in complex tasks such as determining the cause of loss for insurance coverage.

Human involvement remains. Complex cases that the agent cannot handle according to defined regulations are escalated to a human. Additionally, all claims rejected by the agent undergo mandatory human review, providing an extra layer of control and eliminating bias.

Results of AI Agent Implementation at Suncorp

While specific quantitative metrics on time and cost reduction have not yet been fully disclosed, Suncorp is already seeing significant improvements. Key results include:

• Accelerated processing. Significant reduction in claims settlement time due to automation of routine operations.
• Increased accuracy. Achieving 99% accuracy in complex tasks such as determining loss causes and coverage verification, minimizing errors and risks.
• Resource optimization. Freeing up employees from routine tasks, allowing them to focus on more complex cases and customer interactions.
• Scalability. The system's ability to quickly adapt to changes in claim volume, especially during peak load periods.

These results not only enhance Suncorp's operational efficiency but also improve the customer experience, which is a key factor in competitiveness within the insurance industry.

How to Implement This in Your Company

Suncorp's case demonstrates how AI agents can transform complex and multi-stage processes. If claims or inquiry processing in your company is time-consuming and requires checking many parameters, you can adopt this experience:

• Decompose the process. Break down a complex process into many small, logically complete "sub-processes." A specialized AI agent can be developed for each such sub-process.
• Start with narrow tasks. Implement agents in stages, starting with the least risky and most labor-intensive operations where the benefit is most quickly visible.
• Ensure control and monitoring. Create a system that tracks agent performance, accuracy, and identifies anomalies.
• Maintain human involvement. AI agents should be assistants, not complete replacements for humans. Leave complex, non-standard cases, as well as final review and approval, to employees.

If this case sounds like what's happening in your company, our manager can help: he'll analyze your business and niche for free and point out where an AI agent would bring a real result in your case. Message the manager