With AI, insurance claims will soon be processed in real time: PolicyBazaar's Ashish Gupta

With AI, insurance claims will soon be processed in real time: PolicyBazaar's Ashish Gupta
Ashish Gupta
3 Sep, 2018

Gurgaon-based ETechAces Marketing & Consulting Pvt. Ltd, the Softbank Group Corp-backed company that owns online insurance services provider PolicyBazaar, ventured into the online healthcare services market recently with the launch of DocPrime. Ashish Gupta, chief technology officer at PolicyBazaar and CEO at DocPrime, spoke to TechCircle about how the company is using emerging technologies such as artificial intelligence and analytics to spin out new products and change the way it does business. Edited excerpts:

What emerging technologies are you using and why?

We use a variety of emerging technologies ranging from analytics to artificial intelligence to machine learning to blockchain. For us three things matter the most. First, how fast we can serve the customer. Second, the manner in which we respond to the customer and whether that is creating a better experience for them. Third, drive operational efficiency through smart management and bring down costs.

How are you using such technologies to respond faster to the customer?

Earlier, as in any company, a customer would call the PolicyBazaar customer care helpline and navigate a frustrating range of IVR options before getting to an agent who could solve their problem. The customers no longer need to do that. 

We use Amazon Web Services’ (AWS) Polly voice service to offer an interactive voice bot service called PBee Assist. It asks the caller about his or her problem then directly connects him and her to the agent, if required. 

The voice bot is programmed to take care of simple tasks that we often get as requests. The customer doesn’t need to wait in a call queue to get a soft copy of his policy or ask for a change of certain things or verify policy related things. PBee Assist can take care of it. It is also intelligent enough to transfer the call to an agent if it cannot find the solution to a customer’s problem. It will soon be available in Hindi, too, as Polly has been updated. 

Using PBee Assist has led to 15% of cases being closed without intervention and 75% successful identification of customer issues.

We have also experimented with a Polly-based outbound service and have seen success. More than half of the customers waited for the voice bot to deliver the full message before disconnecting the call. 

We have other chatbots, too. Some of our customers said they wanted to chat online instead of calling people. But think about this, if actual human beings chat, then it takes a lot of time, effort and money. We wanted to automate this process and thus we launched a series of chatbots that are intent-driven. These bots, too, can take care of simple tasks. In this way, we are live 24x7.

How have you used technology to make the processing of insurance claims more efficient?

In our vehicular insurance space, we have automated the entire claim verification and disbursal process to cut down on time. Earlier, customers used to send us photos and videos, which had to be seen and verified by an agent. That would take at least 15 minutes and each video would be seven minutes long. Then the paperwork and disbursal would take some time. So, we ended up taking three days for the entire process. 

We automated it to a certain extent first, which enabled us to cut down the process to three hours. But now, with the help of technology, we have brought it down to 30 minutes. 

Soon, it will be real time. We are able to do all this because, instead of an agent, we now use an AWS service called Rekognition that has machine learning built-in and we have tuned it to identify dents and scratches in vehicles from images and videos. The user uploads the pictures and videos and the rest is taken care of by our engines.  

How is analytics for business intelligence helping you to sell your products better?

We are using analytics not just to come up with newer ways of targeted advertising or marketing but also to spin out new products. We’ve used data and artificial intelligence to come up out with a new product called OPD insurance under DocPrime. 

OPD insurance is currently very expensive, priced at Rs 6,000 per person because there aren’t many ways to detect fraud. We are thinking of launching a OPD insurance product in a month’s time at half that price and then further bring it down to Rs 1,500 for a family. 
What we intend to do is use a face detection engine. All the user has to do is take a selfie with the doctor to verify that he is sick and that he has been to an OPD. This will allow us to process the claim faster. 

Also, it will help us create social marketing schemes where the user can post the selfie on social media with a referral link. If a new user joins using that link, then they earn credits.

Has the cost of customer acquisition reduced due to the use of analytics?

We run analytics on data that we collect from Google and from telecom companies, for example, Vodafone. We also track the data of customers who visit the website and check certain products. This helps us understand the customer and his needs better. 

Analytics also help us to modify our products to suit the customer better. We will soon have 30 exclusive products that are not available in the market right now. We also use chatbots in the sales assist format. Once a customer visits the website, he can chat with the bot to find his desired product, following which an actual adviser steps in. 

These bots have changed our sales conversion statistics from 7% to 50%. In terms of advertising, we use Google AdWords and Facebook digital marketing. Data insights have helped us reach a marketing spend to revenue ratio of 280% from just 110% in three years.

You mentioned blockchain. Tell us how you are using that technology.

We have been dabbling with blockchain for some time and are conducting a proof-of-concept with one of our partners. However, I cannot divulge too much. I can tell you that it will be on the payments and data side and enable us to ensure that data is not tampered with by either the policy holder or us and payments reach the right person at the right time.

What kind of investments are you making towards these newer technologies?

While I can’t give you a number, I can certainly tell you that there are at least 30 people continuously working with these new technologies to improve our product offerings and spruce up our operational efficiency. These 30 members are part of PB Labs, which has been in existence for three years and they are divided into data scientists and engineers.