MakeMyTrip is India’s leading online travel company and also owns Goibibo and Redbus travel apps. To date, over 62 million people have planned and booked their trips through these apps. Three of his 10 domestic travelers book flights through the MakeMyTrip suite of apps.
MakeMyTrip today announced that it is collaborating with Microsoft to introduce voice-assisted booking in Indian languages on its platform powered by Azure OpenAI service. The company says generative AI will make the platform more comprehensive and accessible to more customers across the country.
microsoft story india We interviewed Sanjay Mohan, CTO of MakeMyTrip, to explore the potential for generative AI to transform people’s travel plans and help MakeMyTrip better serve its customers.
Here’s an edited excerpt from the interview:
How do you think generative AI will transform the travel industry?
Let’s take a step back and understand what generative AI is good at, especially large-scale language models. The first opportunity is to summarize content from large databases. The entire travel industry is based on a huge corpus of travel data, and making sense of it can be a daunting task.
The second is conversational ability. For example, vacation planning is a nuanced purchase that requires several things to be put together to make up a package and every customer has their own preferences. By doing so, you can elicit the right response from your customers and better understand their preferences. Your customers will also have a better experience if you can use conversations as a means of engagement on your platform.
The third thing that I find most interesting is adding voice to this whole flow, especially Indian languages. We can make travel services more accessible. This will open the industry to a new audience – India’s next 100-200 million users. Although they are not very proficient in English, they are fluent in their own language and are more comfortable communicating by voice than navigating complex apps on their phones. It also improves accessibility for users with disabilities who cannot use the app effectively.
Can you give specific examples of how generative AI can improve the customer experience on the MakeMyTrip platform?
MakeMyTrip has a huge corpus of user-generated content, including hotel reviews for travelers of all categories. Everyone’s travel context is different and they’re looking for something specific in a hotel. Families with children may look for hotels with swimming pools and playgrounds with activities for children. Honeymoon couples might seek out an infinity pool. Solo travelers may be looking for something adventurous. Generative AI helps provide contextual summaries of user-generated reviews. We are experimenting with providing summaries of reviews from other guests who have stayed at hotels that are relevant to the user’s context to help with purchasing decisions.
We’re building conversations that put people at ease before they do their holiday shopping. It’s very comprehensive for those nervous about interacting on the app, such as the type of hotel they’re looking for, the activity they want to do, the type of meal they want to eat, etc. Once all these considerations have been sorted out by conversation, we can suggest several vacation packages that best suit the needs of this particular customer.
Finally, we plan to follow other Indian languages as well, starting with booking flights and holidays in English and Hindi.
How will generative AI impact the role of human travel agents and customer service representatives in the booking and planning process on the MakeMyTrip platform? How are you going to deliver?
We see these chatbots as intelligent adjuncts to our human agents. A basic set of questions that agents normally ask are now asked by these chatbots to identify prospects and assess purchasing intent. For example, chatbots can answer questions about the weather in a city, when is the best time to visit a particular place, or questions about visa requirements.
In some cases, we recommend products to customers for immediate purchase on the spot. And it happens with our app too. If one of our packages fits the customer’s prerequisites, it will result. But some holiday season purchases are so subtle that people want to customize them quite a bit. They want to talk to a human agent to clear up their doubts or ask for more information before making a final purchase. That’s where conversational bots come in handy.
The way I see it, and the way we’re building it, is that these chatbots will be very smart assistants for vacation professionals. As a result, we believe human agents will be significantly more productive and efficient.
What measures does MakeMyTrip take to protect customer data and ensure the privacy and security of information collected and used by its generative AI algorithms?
The data collected by these chatbot interfaces stays entirely within the system, strictly on an “as needed” basis, just like data collected by apps. Therefore, the privacy and security guardrails we employ in our regular apps apply here as well. Collected data is as secure as data voluntarily provided in the app’s shopping flow.
Why did MakeMyTrip choose Microsoft Azure as the platform for its generative AI solution?
After considering several options available today, we chose Azure because of the richness of Azure Cognitive Services, including speech-to-text, text-to-speech, and rich language capabilities. To provide our customers with a smooth and seamless experience, we have chosen to partner with partners who have the most comprehensive set of services for building end-to-end capabilities. Azure Cognitive Services fits the bill perfectly.
Additionally, working with a trusted partner like Microsoft has its advantages. Microsoft adds another layer of assurance with its “responsible AI” mindset.
What are the potential challenges and limitations of implementing generative AI in the travel industry, and how does MakeMyTrip overcome them?
This is still a very early technology, and I believe there are three challenges at the moment.
Given the kind of industry we are in, people want to see pictures of the hotel they plan to stay in and videos of the places they plan to visit. A voice or text interface alone is not enough. You need a multimodal chatbot interface that I don’t think anyone in the industry has yet figured out. This is a tough problem from a user interaction perspective, and we’re using the first few releases to learn what works best for our customers in a series of controlled experiments.
Conversational chatbots need improved voice support for languages to match the colloquial usage of languages spoken or understood by customers in small towns and villages in India. Some experimentation with colloquial and Hinglish usage also helps fine-tune interactions.
And finally, like everything else we release, we want the performance characteristics of the feature to be world-class before we can roll it out to a 100% customer base. New features should work well for all customers. This is the same thing we do with every new feature we release. Again, we’re just following our standard rollout process.
What do you see as the future possibilities for leveraging generative AI to create innovative travel solutions?
The pace of advances in generative AI is so rapid that I think we’ll be talking in quarters or months rather than years.
These are still early days, but my vision is to have voice interactions in all the common languages people speak in India. increase. Anyone in India can use our platform.
This may also allow people with limited manual dexterity, vision, reading and writing, etc. to access the Service, allowing them to interact in the mode that works best for them.
