ICICI Lombard bets big on AI to improve customer experience

The Covid-19 pandemic is an unprecedented event that has taken the world completely by surprise. While it has severely impacted economic activity and put tremendous pressure on healthcare and national resources, it has also presented organizations with an opportunity to relook at their operating model and in many ways define new ways of working.

Girish Nayak, Chief- Service, Operations, Technology, ICICI Lombard General Insurance, explains, “Newer technologies on the digital front, including Artificial Intelligence and Machine Learning will play a major role in providing innovative solutions from a customer experience standpoint. As consumer expectations and behaviours change, we believe that the insurance industry will continue to change by embracing newer technology, be it through an AI chatbot that acts as a Virtual Insurance Advisor, or an AI detection model that acts as a Digital Claims Adjuster or a ML model that acts a smart underwriter. There will be prolific use of AI, ML and cognitive services that will help address needs such as digital adoption, customer experience management, operational efficiency, underwriting profitability, claims optimization and much more.”

Today, agents and channel partners of ICICI Lombard are interacting with customers over video calls, similarly cheque payments have reduced drastically and digital payments have become the norm. There is a complete paradigm shift towards digital means of engagement not only externally but internally as well. A lot of the company’s backend processes that were paper-intensive have been digitized.

ICICI Lombard has a major focus on AI, and the company is exploring how it can use the technology for automating most of its core processes. Explains Nayak, “Over the last few years, we have invested significantly in building up AI and ML capabilities both in terms of infrastructure and capabilities. For example, almost 2 years ago, we had deployed our break-in solution for motor policies that comes up for renewal past its due date, a surveyor comes and checks the vehicle for any damages. This process is now AI enabled. The vehicle owner can take a few photographs and upload it for the AI algorithm to check for any damage. In case the vehicle has dents, spots, any kind of damage – then the AI engine decides, based on the photographs whether the policy is eligible for renewal or not. This renewal process, when done manually used to take about 72 hours depending on the availability of the customer and surveyor. Now, it happens almost instantly and at the time the customer chooses. If the motor renewal case has not qualified to be handled by AI, a detailed virtual survey is subsequently conducted by the surveyor. This gives the customer flexibility of choosing when and where to renew their lapsed motor vehicle policies. As this solution can now identify motor damages with good accuracy, we are taking this solution to the next level for processing claims for low value damages. We are currently testing these in parallel with motor surveyors and testing the accuracy of these models.”

One of the most important ingredients for building quality and large-scale AI and ML models is data. Given the company’s emphasis on data collection and usage over the years, today, it has access to a large amount of data generated by its systems over these years. The company believes that it can use the insights drawn from this data to its advantage in key areas such as underwriting, claims servicing and fraud detection. Explains Nayak, “We are using AI and ML in fraud detection. We have been using AI and ML based fraud detection models that use extensive amounts of data to predict and highlight probable fraudulent claims. Large quantities of internal and external data is processed at the time that a claim is intimated and logged within the system. Since these models follow the continuous self-learning approach, it also helps us to implement solutions that auto-correct very quickly; reducing the time for learning and execution. Similarly, we are currently testing models in the underwriting process that will help us in increased segmentation.”

Similarly on the health side, for corporate health claims, when a customer goes in the hospital for cashless authorization, the company is able to process authorization for 70% of the cases through its AI solution reducing the time for authorization from 90 minutes to 90 seconds. In this case, the policy related information, doctor’s diagnosis and the course of action recommended by the doctor is ingested in the AI algorithm, which decides the admissibility of the case. Earlier, this was decided by a doctor. Based on the case admissibility, an ML program decides on the optimum claim amount to be sanctioned based on the overall policy sum insured and other parameters. The ML program sends a message to the effect to the hospital. This entire initial sanctioning process takes about 90 seconds. The same process when done manually would take a couple of hours. This also frees up the time for doctors to process more complex cases.

Girish Nayak, believes that there are endless possibilities with respect to using AI. Says he, “We expect to see increased usage of AI and ML in in the insurance industry and this should help address needs across policy purchase, customer service, operations and claims optimization to name a few. Utilization of cloud platforms and services will also increase driven by the increased emphasis on big data, AI and ML. As more applications will start moving to the cloud, data security and monitoring will continue to take precedence.”