Generative AI can make travel services more accessible to millions of Indians: Sanjay Mohan, Group CTO, MakeMyTrip
MakeMyTrip is India’s leading online travel company that also owns the Goibibo and Redbus travel apps. To date, more than 62 million people have planned and booked their travel from these apps. Three out of 10 domestic travelers book their flight tickets via MakeMyTrip’s group of apps.
Earlier today, MakeMyTrip announced its collaboration with Microsoft for introducing voice-assisted booking in Indian languages on its platform powered by Azure OpenAI Service. The company says generative AI will make their platform more inclusive and accessible to more customers across the country.
Microsoft Stories India caught up with Sanjay Mohan, MakeMyTrip’s CTO, to find out how generative AI has the potential to transform how people plan their travel and enable MakeMyTrip to serve its customers better.
Edited excerpts from the interview follow:
How do you envision generative AI transforming the travel industry?
Let’s backtrack a little bit and understand what generative AI is good at, especially with large language models. The first opportunity is summarizing content from a large database. The entire travel industry sits on a huge corpus of travel data, and it becomes a daunting task of making sense of it.
The second is conversational capabilities. Holiday planning, for instance, is a nuanced purchase where several things must come together to compose a package and every customer has their own preferences. By making the flow conversational with generative AI, we can elicit the right responses from customers and better understand their preferences. Customers too will have a better experience if they are able to use conversation as the medium of engagement on the platform.
The third one, which I find is the most interesting, is adding voice to this entire flow, especially Indian languages. It can make travel services more accessible. It will open the industry to new audiences – the next 100-200 million users in India – who are not very comfortable with English but are fluent in their language, and who are more comfortable interacting with voice than navigating a complex app on their phone. It will also improve accessibility for users with disabilities who are unable to use an app effectively.
Can you provide specific examples of how generative AI will enhance customer experience on MakeMyTrip’s platform?
At MakeMyTrip, we have a huge corpus of user-generated content, like hotel reviews for every category of traveler. Everyone has a different context for travel, and they are all looking for something specific in a hotel. A family with kids might look for a hotel with a swimming pool and play area with activities for kids, a couple on their honeymoon might look for an infinity pool or a solo traveler might be looking for adventurous things to do. Generative AI can help us provide contextual summarization of user-generated reviews. We’re experimenting with providing summaries of reviews from other customers who have stayed at those hotels that are relevant to a user’s context to help them with their purchase decision.
We’re building conversational capabilities to put people at ease before they make a holiday purchase. It’s very inclusive for people who are nervous about interacting on an app and they can get a lot of their questions answered while providing us with context, such as the kind of hotels they are looking for, the activities they’d like to do, the kind of meals they want to have and so on. Once all these considerations are sorted out via the conversation, we can suggest a few holiday packages that best suit the needs of this particular customer.
Finally, we’re experimenting with voice too, starting with booking flights and holidays in English and Hindi with other Indian languages to follow.
How will generative AI impact the role of human travel agents or customer service representatives in the booking and planning process on MakeMyTrip’s platform? How do you intend to provide a seamless experience that combines AI with human expertise?
We look at these chatbots as an intelligent aide to our human agents. The basic set of questions that an agent would typically ask will now be asked by these chatbots to qualify a lead and assess the purchase intent. For example, questions like weather in a city, a good time to visit a certain place or queries on visa requirements can be answered by the chatbot.
In some cases, it will recommend a product to the customer to purchase right then and there. And that happens on our app too, if one of our packages fit the customer’s prerequisites that’s the result that comes up. But some holiday purchases are very nuanced, and people want to customize it quite a bit. They want to talk to a human agent to clarify their doubts or seek more information before they make the final purchase. That is where the conversational bot comes in handy.
The way I see it, and the way we are building it, is that these chatbots will be very smart assistants for our holiday experts so that they will get more qualified leads that they can close better. We believe the productivity and efficiency of our human agents will get a significant boost as a result.
What measures is MakeMyTrip taking to safeguard customer data and ensure the privacy and security of information collected and utilized by generative AI algorithms?
The data collected by these chatbot interfaces stay completely within our systems, just like the data that we collect in our apps, which is strictly “need-only.” Therefore, the privacy and security guardrails that we employ in our usual apps applies here as well. The collected data is just as safe as the data that one would volunteer on our app’s shopping flows.
Why did MakeMyTrip choose Microsoft Azure as the platform for its generative AI solutions?
We looked at several options that are out there today, and we chose Azure because of the entire bouquet of services on Azure Cognitive Services – be it voice-to-text, text-to-voice or rich vernacular language capability. In order to provide a smooth, seamless experience to our customers, we elected to partner with the one that had the most comprehensive set of services to build an end-to-end capability. Azure Cognitive Services fit the bill completely.
Additionally, there are benefits to collaborating with a trusted partner like Microsoft, who add yet another layer of assurance with their “responsible AI” mindset.
What are the potential challenges or limitations of implementing generative AI in the travel industry, and how is MakeMyTrip working to overcome them?
It’s still a very nascent technology and I see three challenges at the moment.
Given the kind of industry we are in, people want to see photos of the hotels they plan to stay at or videos of the places they plan to visit. Voice or text interfaces alone will not suffice. We need a multimodal chatbot interface that I don’t think anyone in the industry has been able to figure out yet. It’s a hard problem from a user interaction perspective and we’re using the initial few launches to learn what works best for our customers, in a controlled set of experiments.
With conversational chatbots, voice support for languages needs to be refined for the colloquial usage of the language, something that a customer in a small town or village in India would speak or understand. Some experiments with colloquial and Hinglish kind of usage will also help us fine-tune the interaction.
And finally, as we do for every other feature that we launch, the performance characteristics of the feature have to be world-class before we roll this out to 100 percent of our customer base. The new feature has to work well for all our customers. This is no different from what we do with every new feature we launch – we would just be following our standard roll out process here as well.
Looking ahead, what possibilities do you see for leveraging generative AI to create innovative travel solutions?
The pace of progress in generative AI is really fast, so I’m assuming that we will not be talking in years, but quarters or months.
These are still early days, but my vision is to have voice interactions in all popular languages that people speak in India. And if it’s colloquial, it makes services accessible to more people. Anyone in India would be able to use our platform.
This also has the potential to make services accessible for people who have limitations with manual dexterity, vision, literacy etc. and will enable them to interact in the mode that works best for them.