AI: the moment of truth – Kleptika featured in Premium Insurance Magazine

Premium Magazine AI Thierry Petrens Kleptika CEO

Economic pressure has driven organisations to make archaic processes more efficient by minimising the delays and guaranteeing information security. Artificial intelligence is the key, explains Thierry Petrens.

Read more:

Artificial Intelligence (AI) has moved away from dreams and science-fiction to become a reality, showing high positive impact in our everyday life in many different sectors. Latest technology innovations and new customers’ demands have accelerated the AI process of adoption and its speed of propagation. The insurance sector is realizing the need of this game-changing concept to transform in depth the relations between insurance players and their customers.

The market opportunity for business applications using AI will reach more than USD10 billion by 2024, making AI one of the fastest-growing market segments. The drivers for the adoption of AI are multiple, but three trends are emerging: first, the adoption of Big Data together with the explosion of multi-format collected data require finding cost-effective solutions to efficiently make smart decisions. Second, recent economic pressure has driven organisations to make archaic processes more efficient by minimising the delays and guaranteeing information security in an environment where less human resources are operating. And third, the power needed in terms of AI computation and the ability to follow higher and higher demand has only been reached recently, mostly within cloud architecture.

More specific to the insurance sector, the customers’ behavior and the new communication technologies have changed the traditional landscape. Understanding these changes and being able to respond to the new operational and functional needs are too complex to be handled manually, and thus require relying on automated learning rules and patterns from data using Machine Learning approaches.

Also, the new generations of users are more empowered: they were born within a digital environment, they demand high-level customer experience while considering quality as a standard, they use vocal interface, comment quickly on social media, want to be in control while demanding transparency and security, and are supported by strong regulations about privacy and traceability. They are also massively using mobile and cloud to access their professional and personal life on their smartphones, anywhere and at any time. Last, they are impatient; they live in an era of instant gratification brought by digital vehicles such as Facebook or Instagram.

Insurance carriers can definitely benefit from these new technologies to answer their customers’ demands. Recent implementations have shown dramatic improvements in adding AI solutions to solving key problems such as document management, knowing your customer (KYC), flexible pricing, fraud detection and claim handling.

Documents Intelligent Processing (DIP)

Latest innovations in technology and adoption of smart mobile devices have changed in depth the way documents are captured, read and analysed. Since tablets and smartphones have cameras, they become part of the document management process, at the same level of smart and connected scanners.

Lots of mobile apps have scanning capabilities, only a very few of them offer a complete set of features needed to complete the full document management process and at the same time improve the customer experience. Capturing the image itself hasn’t changed so much; it’s the adjacent tasks performed at the time of capture that make the difference.

AI concepts and mostly Deep Neural Networks (DNN) along with piloted scanning, document classification, automatic data extraction and data validation are done today almost entirely at the device level, with the absence of or very little human intervention.

Integration of a workflow tool allows to control the number and type of documents that need to be captured, while certification of the capture by adding time stamp, geo-localisation data and deep watermark security tags make the document a legal evidence.

DNN ensures more than 99 percent of automated recognition of complex documents, such as handwritten ones, mobile snaps, barcodes, multi-formats, multi-language, low quality pictures, unstructured and unconstrained documents.

For example, after a car accident, the adoption of smart mobile capture and DNN analysis allows to capture on the spot all the necessary documents: identity of the different drivers, insurance policies of vehicles, pictures of the damages. But it also validates in real time the eligibility of the premiums, the sending of assistance if required, and the connection with the authorities (police, insurance companies).

The information present in the needed document is most of the time very unstructured. Different insurance certificates from different companies will have the policy number in different places. This number will have different formats, font and structure.

From an AI point of view, only the presence of the policy number matters, regardless of the location on the document. Using complex logic and initial training, AI can isolate the required information and transfer the data to the next process task.

Integrating all AI and DNN features into the claim processing may easily reduce by 70 percent the overall handling time, reduce financial costs and dramatically increase customer satisfaction.

Knowing your customers (KYC)

One key area of success is related to elevating the customer experience by understanding deeply what, when and who needs a specific service or product, thus reducing churn ratio.

BI already takes advantage of the data collected during customers’ interaction through the different platforms of communication, to offer personalised content and a first level of cross- and up-selling offers.

Adding telematics via behavioural intelligence to collect members’ lifestyle and driving patterns completes their profile in a highly personalised way, delivering proactive services and communication.

Healthcare members may have their condition monitored in real time by connecting the behavioural platform to their medical or smart devices, and receive notifications, alerts or even medical resource call based on their semantic time, place and segment.

Drivers may benefit from personalised premium by sharing their driving behaviour, or receive assistance in case of real-time crash detection or abnormal situation within their usual behavior.

Behavioural intelligence platforms based on mobile sensors offer today 40 percent more accuracy than hardware devices attached to a vehicle, and may use the same medium to exchange voice, video and chat.

Multimedia chatbot using Natural Language Processing (NLP)

The most used medium by the millennial generation today is chat because it is the most direct and natural way to interact using voice, text or video. Natural Language Processing but also emotional analysis are key to address customer requests and wishes in a personalised way.

Setting up a true channel of interactions through this medium requires to be ready to support the customers in every language and dialect, in a NLP mode and not only via keywords.

NLP brings the level of interaction to the one users can experience with a human agent, and allows them to express their request with their own words and in a complex way. By example, “I want a family insurance” will be translated by a premium for the husband, the wife and the kids, and the customer profile will indicate that this premium needs to cover international trips to the member’s homeland.

The channel should be open 24/7, propose almost the same services as the other channels, in real time or almost real time, support interaction and proactively anticipate the needs. A seamless and agile communication with the backend systems of the organization is key to success.

Evolution versus revolution

Artificial Intelligence, including behavioural intelligence, secured and proven capture of documents, automatic reading, classification and extraction of data independently from language and formats, supported by attended RPA and DMS, can only grow in importance and capability.

From medical to insurance and finance, the need of AI is clearly understood by worldwide players. Successful implementation requires deep business understanding as well as technical knowhow, but efficiency and acclaimed CX do need the total understanding of customers’ needs.

The return on Investment is very attractive, even when paying a premium for truly effective solutions and advanced services.

Finally, leaving a large part of the processing to Artificial Intelligence will free the staff from unnecessary manual tasks and validations, and develop human intelligent interactions.

This is not a revolution, this is a natural evolution.