Could robots become the next travel retailers?

Landry Holi

Head of Innovation - Engagement & Governance, Amadeus IT Group

This content is only available in this language.

In the beginning, travel was sold through brick and mortar travel agencies. Then came the web, followed by mobile. Now, with further proliferation of connected devices and channels (wearables, social networks, chat, gaming, interactive TV, connected cars…), there are many more opportunities for travel retailers to connect with travellers.


With this proliferation, automation (with a human touch!) becomes critical to scale such opportunities. In fact, we’ve already blogged about the chat bot phenomenonand how it’s supporting the growing trend of chat channels being used for searching, booking, and paying for travel.

But in the context of an omni-channel strategy, what role could robots play in the offline travel retail stores of the future?

To start experimenting with the applications of robotics to offline travel retail stores, we recently welcomed a new member to the Amadeus Innovation & Research organisation: 1A-TA. We first presented this little robotic experiment with 1A-TA at Innorobo(24-26 May 2016, Paris), an event showcasing state of the art robotics.

Meet 1A-TA, the travel retailer's assistant

1A-TA is actually Pepper, a humanoid robot designed by Softbank Roboticsto interact with human beings. The first use case we implemented is quite simple:

You enter the travel agency and the travel consultant is busy. You sense that you will be waiting for a long time before you can get some recommendations on where to go for your next holiday. Should you stay or should you go?

Suddenly, you hear a friendly voice from a cute little robot that invites you to play a game: “Hello, I will show you some pictures and you can tell me which ones you like. If you like this picture, you can smile, say it out loud, or just press “I like” on my tablet. Do you like coffee? The sea? Traffic jams?”

With your permission, 1A-TA is progressively creating your traveller profile. The robot extracts the characteristics of your personality from random pictures (i.e. not necessarily destination related), then matches them with a list of relevant destination profiles.

Intelligence Algorithms

Behind the scenes, in addition to Softbank’s Pepper cognitive development kit, we integrated some complex deep learning artificial Intelligence algorithms. These algorithms were trained through billions of news articles enabling the robot to link images to keywords themselves that were transformed at some stage into numerical n-dimensional vectors. This enables the robot to link your profile, which it created through your dialogue, with destination profiles, thanks to ‘simple’ mathematical functions.

After displaying a shortlist of destinations, 1A-TA proposes the most relevant one. 1A-TA then transfers this information automatically onto the travel consultant’s desktop, along with all profiling data collected through the exchange. The travel consultant is now ready to add their value by refining your destination choices and proposing inspiring relevant ancillaries. Broadly speaking, they can simply focus on providing key added-value recommendations that only a human, and not a robot, can provide you with in order to make your trip truly unique.

So here, we deliver the best of both worlds, offline and online. 1A-TA is able to capture a lot of data about the traveller (as we do online), yet the experience is still engaging (as we can achieve offline). And above all, travel consultants can leverage this to maximise the value of their role.

From a first prototype, where do we go next?

During the Innorobo event, we received some very positive feedback. I would even say that we sensed a good product and market fit from the interactions we had.

Nevertheless, even though we managed to gather such promising feedback, we must remain humble and realistic. In fact, we have pinpointed some technical uncertainties that would need to be resolved in order to make our experiment something that could be deployed in the ‘real world’. To name a few: deep learning profiling algorithm accuracy and cognitive capability extensions of the robot. Interestingly enough, a lot of these improvements would benefit other channels we have also started experimenting with, such as chat bots, interactive TV, and connected cars.

The next step for us, is to experiment and refine use cases within existing travel retail shops. Despite the fact that we know we will face further challenges in different areas (technology, business, legal, social…), we are very excited to contribute to shaping the future of travel!

So, don’t be surprised if you run into 1A-TA the next time you visit your travel agent!


Travel Platform