The Indian airline IndiGo, with Pieter Elbers at the helm since October 2022, has launched a chatbot that can answer questions in different languages. Competitor Air India has also launched a bot where only 15% of questions have to be forwarded to a live agent, Air India said.

 

IndiGo wants to significantly reduce the workload of customer service employees with the latest technology. IndiGo's digital team worked with Microsoft to develop the chatbot. The result is what IndiGo describes as a multilingual “unique ticket booking platform across the network”. The chatbot has the name 6Eskai received, according to IndiGo.

6Eskai

6Eskai works on the basis of 1,7 billion parameters, which can be used to find solutions for various customer questions. The bot is programmed using extensive prompt engineering to mimic human behavior, recognize emotions and even put a humorous spin on interactions. According to the company, the workload of contact center agents has been reduced by 75% after the 'soft launch' of the chatbot. The chatbot would be ideal for booking tickets, including applying discounts, booking add-ons, performing online check-ins, helping with seat selection and trip planning.

Maharaja knows almost everything

With the launch of 6Eskai, IndiGo is following in the footsteps of Air India, which also introduced an AI-powered chatbot earlier this month. This virtual agent, named Maharaja, is also multilingual: the bot knows Hindi, English, German and French, among others. Maharaja is said to answer more than 6.000 questions every day within 80 seconds; Only 15% of queries require resolution by a 'human' Air India contact center agent, the company said.

Maharaja can answer customer questions in 1300 areas ranging from flight status, baggage allowance, packing restrictions, check-in, frequent flyer awards, airport lounge access, flight changes and refunds, according to Satya Ramaswamy, chief digital and technology officer of Air India, in the press release.

Learning from customer contacts and making adjustments

Maharaja was launched in March this year, but wasn't doing very well at the time, says Ramaswamy. Lacking sufficient data to train the Gen AI algorithm, the company used contact center data to power the initial learning phase. On the one hand, Maharaja had to learn in real-time from customer interactions and on the other hand, manual adjustments had to be made to prevent problems such as AI hallucination. Adding speech recognition from the Microsoft Azure Gen AI environment has also quickly improved the accuracy of answers. The development team actively communicated about this learning period, says Ramaswamy. That team consisted of prompt engineers, data scientists, natural language processing experts, conversational designers, copywriters, developers, cloud experts and user interface teams.

Enterprise AI

However, Air India is taking GenAI a step further and has also built an “enterprise Gen AI-powered chatbot” that is integrated into Microsoft Teams. This enterprise bot helps senior executives collect real-time operational information. A similar system is being developed for pilots and cabin crew to better access the extensive flight manuals. (IndiGo, Air India)

Featured, Technology