The Art of Possible

Create intelligent document processes at AXA with AI

AXA has about 108+ million clients who on consuming a medical service, the members submit their claims to the insurer. Claims are received through various channels including mobile apps.

At the insurer end, claims processing is a key operational activity. There are scanned receipts received with a claim which are typically embedded or attached in the claim document. It requires manual intervention to extract and validate data and manual interventions pose the usual challenges like adding delays in the claim processing, manual errors and additional auditing needs. Processing claims involves reading the claims form, invoices, policy document and the underwriting authorization letter. Most of the process can be extensively automated using RPA and other backend workflow integrations.

Given the volume of claims to be processed and the need for the manual processing, the insurer ends up having a large operations team for claims processing.

The need is to have an automated mechanism for data extraction and validation from the claim documents to reduce the manual effort.

The system should be able to perform straight through processing and in case of accuracy issues should be able to learn from human correction i.e. have the capability of Machine Learning based learning.

Enable business innovation at AXA with AI-led IT operations

As one of the largest global insurers and being a large corporation AXA has a different set of IT challenges with digital transformation being one of the key IT initiatives. With hundreds of applications within the organization and it undergoing digital transformation, maintaining business continuity is an uphill task. While the common application stack is maintained by AXA Go, the situation will not be too different across geographies. This has the potential to impact customer experience adversely, if not handled efficiently. In such scenarios, it is observed that the number of tickets related to IT systems (specially for L2 and L3) also increases significantly. IT Support team works round the clock on ticket resolution to reduce noise in production and to meet the SLAs.

In IT Operations scenarios, it is observed that the majority of the tickets have some patterns and their resolutions are also standard.

With a lot of data getting generated in the individual applications as logs for every event there is a lot of historical ticket data available in the system that remains unexplored. It is difficult and challenging to manually go through the data and identify the pattern or co-relations between them to derive insights. Due to the lack of this analysis, operations teams continue to invest a significant amount of time in analyzing every problem individually.

There should be a system, which consumes the historical application logs and ticket data to create co-relations leveraging AI capabilities and helps in identifying the root cause for commonly occurring failures. There is a need to have an automated way of remediating these tickets and where possible, predicting these failures and also fixing the issues before they occur. If a failure has occurred and automated handling is not possible, for manual ticket handling, the operations team should be able to easily analyse and fix the problem.

Reduce dependency on human resources at AXA with RPA for Testing

In the age of digital transformation, large companies like AXA look at automation as the first and key lever to optimize every area of application maintenance & support operations including testing and validation, using varies tools for performing test automation.

Traditional test automation has some limitations on various fronts, for example, different tools support automating a particular type of application testing like Selenium for web applications and UFT for web and desktop applications. Whereas the RPA approach is platform independent and can be used to automate web, desktop, mainframe based applications and those running on virtual machines like Citrix.

The traditional test automation approach is also costly in the long run and requires high technical expertise to write the test scripts and has limited configurability which is not all encompassing.

The need is to have a comprehensive automated testing capability to handle:

• Life Cycle Test Automation – Business Flow Automation involving multiple technologies and platforms

• Test Data Management – Test data generation involving multiple applications, time intervals, realtime and transactional

• Citrix / Legacy Automation – surface automation image recognition and OCR

• Unstructured Data Processing – Bank statements, letters, cheques and inbound and outbound correspondences

• Non-Core Automation – Reconciliations, Job Monitoring, Metrics Reporting and Data Migrations

• User Acceptance Testing – Business centric test automation

Amplify procurement transformation at AXA with AI

With an increasingly complex ecosystem of technology and market information at AXA, Procurements Leaders have to tackle certain challenges to meet the rising expectations. Procurement is increasingly expected to evolve from a support function to a key strategy enabler. Critical tasks include optimizing cost, bringing in innovation, and moderating risks and costs.

91% of CPOs agree that procurement needs to be more agile to respond to market changes and 78% of Procurement Leaders say that cost reduction is their top priority. From this perspective, large organizations like AXA are looking to make procurement visible, meaningful, and simple by addressing key challenges in a way to make spend management both insights-driven and autonomous. Key challenges include:

• Data spread across multiple ERP systems

• Missing connections between internal and external insights

• Limited Spend Visibility, supply risk and spend leakages

• Lack of insights on performance issues

• Critical obligations in supplier contracts that lack proactive monitoring

• High Contract leakage and non-compliant spend

• Limited ability to influence buying patterns

• Lack of awareness on opportunities that exist

• Business decisions delayed or not the most optimal

• Different views across various stakeholders