MakeMyTrip is India’s leading online travel company that also owns the Goibibo and Redbus travel apps. Thus far, greater 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 corporate says generative AI will make their platform more inclusive and accessible to more customers across the country.

caught up with Sanjay Mohan, MakeMyTrip’s CTO, to learn how generative AI has the potential to remodel how people plan their travel and enable MakeMyTrip to serve its customers higher.

Edited excerpts from the interview follow:

How do you envision generative AI transforming the travel industry?

Let’s backtrack a bit bit and understand what generative AI is nice at, especially with large language models. The primary opportunity is summarizing content from a big database. Your complete travel industry sits on an enormous corpus of travel data, and it becomes a frightening task of creating sense of it.

The second is conversational capabilities. Holiday planning, for example, is a nuanced purchase where several things must come together to compose a package and each customer has their very own preferences. By making the flow conversational with generative AI, we are able to elicit the suitable responses from customers and higher understand their preferences. Customers too could have a greater experience if they’re able to use conversation because the medium of engagement on the platform.

The third one, which I find is probably the most interesting, is adding voice to this whole flow, especially Indian languages. It could make travel services more accessible. It can open the industry to latest audiences – the subsequent 100-200 million users in India – who are usually not very comfortable with English but are fluent of their language, and who’re more comfortable interacting with voice than navigating a posh app on their phone. It can also improve accessibility for users with disabilities who’re unable to make use of an app effectively.

Are you able to provide specific examples of how generative AI will enhance customer experience on MakeMyTrip’s platform?

At MakeMyTrip, we’ve an enormous corpus of user-generated content, like hotel reviews for each category of traveler. Everyone has a distinct context for travel, and so they are all searching for something specific in a hotel. A family with kids might search for a hotel with a swimming pool and play area with activities for teenagers, a pair on their honeymoon might search for an infinity pool or a solo traveler is perhaps searching for adventurous things to do. Generative AI may also help us provide contextual summarization of user-generated reviews. We’re experimenting with providing summaries of reviews from other customers who’ve stayed at those hotels which are relevant to a user’s context to assist them with their purchase decision.

We’re constructing conversational capabilities to place people relaxed before they make a vacation purchase. It’s very inclusive for people who find themselves nervous about interacting on an app and so they can get loads of their questions answered while providing us with context, similar to the type of hotels they’re searching for, the activities they’d wish to do, the type of meals they wish to have and so forth. Once all these considerations are sorted out via the conversation, we are able to suggest a number of 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 support representatives within the booking and planning process on MakeMyTrip’s platform? How do you plan to offer a seamless experience that mixes AI with human expertise?

We have a look at these chatbots as an intelligent aide to our human agents. The essential set of questions that an agent would typically ask will now be asked by these chatbots to qualify a lead and assess the acquisition intent. For instance, questions like weather in a city, a very good time to go to a certain place or queries on visa requirements may be answered by the chatbot.

In some cases, it’s going to recommend a product to the shopper to buy right then and there. And that happens on our app too, if one in all our packages fit the shopper’s prerequisites that’s the result that comes up. But some holiday purchases are very nuanced, and other people wish to customize it quite a bit. They wish to seek advice from a human agent to make clear their doubts or seek more information before they make the ultimate purchase. That’s where the conversational bot is useful.

The way in which I see it, and the way in which we’re constructing it, is that these chatbots shall be very smart assistants for our holiday experts in order that they’ll get more qualified leads that they will close higher. We imagine the productivity and efficiency of our human agents will get a big boost in consequence.

What measures is MakeMyTrip taking to safeguard customer data and make sure the privacy and security of data collected and utilized by generative AI algorithms?

The info collected by these chatbot interfaces stay completely inside our systems, similar to the information that we collect in our apps, which is strictly “need-only.”  Due to this fact, the privacy and security guardrails that we employ in our usual apps applies here as well. The collected data is just as secure as the information that one would volunteer on our app’s shopping flows.

Why did MakeMyTrip select Microsoft Azure because the platform for its generative AI solutions?

We checked out several options which are on the market today, and we selected Azure due to your entire bouquet of services on Azure Cognitive Services – be it voice-to-text, text-to-voice or wealthy vernacular language capability. So as to provide a smooth, seamless experience to our customers, we elected to partner with the one which had probably the most comprehensive set of services to construct an end-to-end capability. Azure Cognitive Services fit the bill completely.

Moreover, there are advantages to collaborating with a trusted partner like Microsoft, who add yet one more layer of assurance with their “responsible AI” mindset.  

What are the potential challenges or limitations of implementing generative AI within the travel industry, and the way is MakeMyTrip working to beat them?

It’s still a really nascent technology and I see three challenges in the meanwhile.

Given the type of industry we’re in, people wish to see photos of the hotels they plan to remain at or videos of the places they plan to go to. Voice or text interfaces alone won’t suffice. We’d like a multimodal chatbot interface that I don’t think anyone within the industry has been capable of determine yet. It’s a tough 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 must 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 type of usage may also help us fine-tune the interaction.

And eventually, as we do for each other feature that we launch, the performance characteristics of the feature need to be world-class before we roll this out to one hundred pc of our customer base. The brand new feature has to work well for all our customers. This is not any different from what we do with every latest feature we launch – we’d just be following our standard roll out process here as well.

Looking ahead, what possibilities do you see for leveraging generative AI to create revolutionary travel solutions?

The pace of progress in generative AI is actually fast, so I’m assuming that we’ll 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 folks speak in India. And if it’s colloquial, it makes services accessible to more people. Anyone in India would have the ability to make use of our platform.

This also has the potential to make services accessible for individuals who have limitations with manual dexterity, vision, literacy etc. and can enable them to interact within the mode that works best for them.