Seema and her team at IFFCO Tokio General Insurance Company are working to make it a lot easier for their customers to deal with vehicle damage claims with an AI-based mobile app, which helps settle claims quickly and easily.

Learn how you can see results with Watson Assistant for Customer Service

How are you using AI?
We use AI to assess the claims of damaged vehicles through image processing. A customer can upload photos of the damage through a mobile app and the AI engine analyzes the photos and, within seconds, generates a list of needed repairable and replaceable parts. These parts are then searched in the historical claims database for the average cost of repairor replacement. The total cost is displayed to the customer on their mobile app within few minutes, and they can accept or reject the offer. If it’s accepted, payment is made through a payment service within a few minutes. The assessment data is also analyzed by a claims officer at the back office for verification purposes. Any discrepancies are sent as feedback to the AI engine to help it learn.

What benefits are you realizing with AI?
We have used the system for about two months and it’s resulted in a much better customer experience. We’re currently using it on about 20 car models, but we will add about 30 more in coming months. We will also extend this service to do pre-inspections of cars where break-ins occur. We have found this process to be truly transformational. The end-to-end claim settlement time has been reduced from three or four hours to just 15 minutes.

What have you learned that you wish you knew when you first started with your work with AI?
Proper data input is essential for machine learning, and continuous data feedback is essential to increase the accuracy of the AI engine.

What advice would you share with others who are considering using AI?
Machine learning requires perseverance. Be prepared with a comprehensive set of input information to the AI engine. Do a thorough proof-of-concept run with vendors before finalizing the product. Test the AI engine with varied input data to ensure accuracy. And run a parallel, human-assisted model to gain confidence in AI.

Interested in more stories of women in AI?
View the 2019 list

Want to learn more about the AI behind this work?

Watson is IBM's suite of enterprise-ready AI services, applications, and tooling.