Fluffy: Your AI Travel Companion

Picture this: It’s early morning, and you’ve just landed in Dubai. It’s your first time visiting a city you’ve never been to, a place that is filled with endless experiences to explore. However, the excitement in the air is still coupled with a sense of uneasiness. How can you make the most of this beautiful city? You can’t help but think about questions like:

1. Where can I enjoy the nightlife in Dubai?
2. What are the top shopping destinations?
3. Where can I get a local SIM card for my phone?
4. What are some common phrases used by locals?
5. Are there any cultural customs in Dubai that I should be aware of?

In today’s digital age, booking a trip is just the first step. What follows is a maze of questions and anxieties about making the most of your time at your new destination. Beyond securing experience tickets and accommodations, guests are seeking reassurance, guidance and an elevated experience. That’s where Fluffy comes in—Headout’s travel assistant designed to ease travel anxieties and provide real-time solutions to your trip-related queries.

The problem statement

Trust in a brand can often outweigh price when it comes to booking online experiences. Users want to know if they are in good hands not only while booking but throughout their journey. This also includes the discovering, planning and post purchase phase of the trip.

While one can pay travel agents to create their trip itineraries, this is usually an expensive option. Planning on your own by relying on word of mouth, blogs and other internet websites can be a really time consuming process. This is where Fluffy comes in. It caters to that category of users that seek reliable, fast and accessible answers to all their travel queries without the hassle of browsing multiple online sources.

When travellers think about their trip and its associated questions- they should think of Headout. Whether they have questions about things to do, hidden gems, local attractions, common phrases or even safety- Fluffy should guide them at every step.

Solution: Introducing Fluffy at your fingertips

As a part of the initial proof of concept, Fluffy was introduced to all those users who had successfully booked one experience with us in Dubai. In the first milestone, the plan was to build it as an app only feature.

Why was this feature exclusive to apps?

Today users can book an experience with Headout both through the website and the app. However, according to the user journey of Headout's guests, the app is where most users engage soon after booking an experience with us. By introducing Fluffy on the app, we wanted to build trust and provide timely help to the users during their travels, making Headout a dependable source for real-time support. This approach not only would enhance the user experience but would also foster retention, encouraging users to keep turning to the Headout app for future trips.

Integrating Sendbird

To test out Fluffy with minimum effort, we integrated Sendbird's AI chatbot, an advanced conversational AI solution designed to integrate seamlessly into apps and platforms, enhancing user engagement and support. It leverages natural language processing (NLP) to enable intuitive, real-time interactions for customer support, onboarding, and community engagement. Its features include multilingual support, sentiment analysis, and customisation options, allowing Headout to create tailored conversational experiences that align with our brand and user needs.

Fluffy's chat interface

We introduced an intermediate introductory screen where we have a pre-defined list of prompts for the user. If the user taps on a prompt or types a question and hits enter, we will take the user to the Sendbird's chat interface.

Segregating prompts according to the lead time of a user

Why does the lead time of an experience matter for Fluffy?
In this experiment, Fluffy was being introduced once the user has successfully completed one booking in Dubai. If the user is a low lead time user i.e. their experience is going to start soon- the assumption would be that they are already travelling in Dubai at the moment. Hence, the kind of prompts we would suggest would be about how they can navigate Dubai easily. Alternatively for high lead time users, we would show prompts which would help them plan their Dubai trip better.

Prompts for low lead time users:
1. Where can I enjoy Dubai's nightlife?
2. What are the top shopping destinations?
3. Are there any offbeat experiences in Dubai?
4. I'm looking for adventure sports.
5. What are the best places to try local cuisine?
6. What should I know about Dubai's culture?
7. Can you teach me commonly used phrases?

Prompts for high lead time users:
1. What are the top attractions in Dubai?
2. Can you recommend fun experiences for my kids?
3. Where can I get a local SIM Card for my phone?
4. How can I get to my hotel from the airport?
5. Tell me what all essentials I should pack.
6. What should I know about Dubai's culture?
7. Can you teach me commonly used phrases?

Loader

Once the user taps on any of the prompts or types a question and hits enter, we load the Sendbird interface. However, loading the interface can take some time, hence we introduced a full page loader between the two screens.

Entry points for the user to access Fluffy

Since this was the first time we were testing out Fluffy, we wanted to ensure maximum visibility. These were the entry points we looked at:

Home page

Along with Fluffy being present in the bottom navigation bar, to draw attention towards it we would show a bubble callout. This callout had the Fluffy mascot animating, along with a list of questions being displayed in the form of a marquee animation. The user can choose to dismiss it manually.

This callout appears only for the very first time the user visits the home page and if the user has not already interacted with Fluffy.

My bookings page

In this page, we show all the upcoming and past bookings to the user. We introduced a callout below the first Dubai booking card in the page. This callout is a contextual nudge for Fluffy, where the user can ask questions about their trip directly. The callout remains within the card till the experience has been completed by the user.

Manage bookings page

This callout was introduced to cater to the in app booking flow. Once the user completed a Dubai booking with us, we take them to the manage booking page to view their booking. To introduce Fluffy early on in the flow for such users, we added Fluffy's callout in the middle of the page.

Post Conversation with Fluffy

Once the user has interacted with Fluffy, which means they have asked a minimum of one question and then they hit back, we display a review pop-up asking them to rate Fluffy.

Positive review

If the user has had a positive experience with us, we thank them for the positive review hoping to see them back soon.

Negative review

In case the user has had a negative experience with Fluffy, we wanted to understand how we could improve the feature through their valuable feedback. Hence we introduced an optional field, where the user can talk about their experience which would further help us improve the feature.

How did it perform?

Overall Fluffy received a lot of love from our users. Not only was the adoption pretty good, we received a lot of positive responses from our users. However, we did not receive a lot of repeat users for Fluffy. Considering that the feature could further improve in providing real time responses, it was rolled back from our apps.

For me, however, the major highlight of this project was the pace with which we executed this entire feature. From design to releasing it in apps, we did it in just 2 weeks.

We could not have pulled this off without the amazing team working on it. Special shoutout to Ramakrishna V. for his support and guidance throughout the project, Roshan Kumar and Zanetta Malar for creating Fluffy's mascot and the introductory video, Raghav Nair for all the product inputs and Rahul Halder for developing the final output.