Send a quick enquiry by e-mail to the insurance company – and in next to no time you will get an answer. That is satisfactory for both customers and companies. However, the mass of e-mails often gets in the way of this. The e-mail bot eRA promises to help. This trained bot at HDFC ERGO uses IBM's Watson.
Bots used to have a bad reputation – impersonal, inflexible. This has now changed. Bots can help us humans in many areas of life. Amongst other things, they can for example answer customer enquiries for companies that come in by e-mail. This saves a lot of time in everyday office life. Especially as many enquiries are repeated and thus offer the employees much potential for frustration. This is exactly where the eRA e-mail bot of HDFC ERGO comes in. ERA is the abbreviation for “e-mail Robotic Automation”.
Already in October 2018, HDFC ERGO launched the bot for customers who preferred to contact them via e-mail. It is a trained bot, which uses Watson from IBM – an artificial intelligence computer program developed specially for chat bots and virtual agents. Records of specified categories were predefined as key factors for the e-mail process. “The bot can answer our customers immediately and offer a solution to the questions asked”, says Mehmood Mansoori, President Shared Services & Online Business at HDFC ERGO General Insurance.
eRA is currently able to process more than 400,000 e-mails per year according to HDFC ERGO. That is 24 percent of the customer enquiries coming in by e-mail. For the financial year 2021, HDFC ERGO plans to be able to process 35 percent of the customer enquiry volume with the bot.
“ERA offers us many benefits”, explains Mehmood Mansoori from HDFC ERGO. “Replying to e-mails has now been reduced from four hours to about five to ten minutes.” The bot is trained to understand the requirements of the customers quickly, put them into context and offer solutions immediately. And that increases customer satisfaction.
An important factor for eRA is the Natural Language Processing (NLP) – an area of artificial intelligence. The technology tries to capture natural language and to process it by computer with the help of algorithms. This should enable computers to understand and interpret human language. “The NLP capability of eRA helps to recognise customer enquiries better and to reduce human error”, says Mehmood Mansoori.
How does eRA work in practice? “IBM Watson is used to categorise data based on specific words and patterns provided by a large number of customers during a request for service”, explains Mehmood Mansoori. “These words establish a relationship and provide the required intentions for a specific category of requests.” As soon as the intention has been identified, eRA executes the predefined series of actions and checks. A corresponding response is subsequently sent to the customer.
A specific example: “If customers ask in an e-mail specifically for a copy of a policy, they use words like: send policy, need deposit, request insurance documents, provide a copy of the policy and many more”, explains Mehmood Mansoori. These words are used to identify the intention as “Policy Copy”. As soon as the system has understood the intention, it will retrieve a copy of the policy. “It will then send an appropriate response to the customer with the copy of the policy”, says Mansoori. If eRA cannot identify the intention or the request is not a known scenario, it is forwarded to the manual E-Mail team.
By using the e-mail bot, HDFC ERGO can for example also process service requests for the status of claims, cancellations or the transfer of insurance policies quickly. The operators say that up to now, the customers have a positive perception of eRA. At least, it is now possible to process requests much faster. “In particular, it has improved our cost effectiveness”, says Mehmood Mansoori from HDFC ERGO.
Text: Benjamin Esche