Previous Next
A sprint in the life of Sabrina
What's next in Tech?
Green Orange wins Golden DIA award in the Data category!
The 7 dimensions of digital maturity you need to know before transforming.
Employee in the spotlight: Bas!

Chatbot hackathon; 3700 lines of code in 48 hours

2017, 04, 10

Vera Engelbertink

We were invited to join the first Pioneers Discover Alpine Tourism Hackathon two weeks ago. The Swiss region of Engadin, St.Moritz organized this hackathon to get startups from various fields to work on innovative solutions in the fields of gamification & new media, data analytics and chatbots to maximize the region’s tourism potential. We were happy to join in this adventure of course and were going to develop a chatbot within 48 hours.

The challenge the region is facing is substantial. The visitors to the region are aging and there is no attraction of a younger target audience. Therefore, there is a need to come up with innovative solutions to be able to attract new target audiences to the region.

We believed the way to face this challenge is to get perspectives from different angles. So we assembled a multidisciplinary team; a developer, consumer psychologist, content marketeer and a machine learning and AI expert. AKA Team Chatbot. Our objective was to build a bot that can support a (future) visitor to the region along his or her whole customer journey; before, during and after the holiday. This way, the chatbot would be an important inspiration and information touchpoint that can assist the visitor in a personalized and easy way.


We believe a good chatbot conversation consists out of a number of building blocks: Start, Intent, Personality, Chitchat, Decisions and Feedback. We start with onboarding the user and managing expectations of what one can expect when engaged in a conversation with chatbot Alpi. Furthermore, we capture the user's’ intent; what is his/her goal of the conversation? Personality is also an important building. By capturing personality you can create the perfect and most enjoyable digital conversation, as you tailer the conversation to the person on the other side. For example is someone an extravert or introvert person? Both would have a personal preference in communicating. Personality is also important in using chitchat in the conversation. Chitchat is what ‘glues’ the conversation together and makes it go more smoothly. An introvert person would have different types of chitchat than an extravert person would for example. Furthermore, asking certain questions to give relevant information and suggestions that lead to a decision are crucial. Last but not least, gaining feedback from the user is very important and valuable for future improvements of the digital conversation.


We aimed, and succeeded, in capturing these building blocks and elements in the digital conversation with Alpi.

Orientation phase:

In this phase we discover whether the visitor is a match with the region of Engadin. During this stage it is important to inspire and inform the (potential) visitor of the region of Engadin. Furthermore, you want to capture elements from the building block of the conversation.

So here we start the conversation and capture the intent of figuring out whether the visitor and the region are a match. A set of questions will give Alpi more information on when the person wants to go on holiday, what kind of holiday they are looking for and with whom they are planning to go. This will also give us insights into personality type, as an extravert person is more likely to go on holiday with a group for example. We also want to capture things that can be useful later on, such as type of activities, in this case hiking. This way, Alpi can be as relevant as possible when the visitor arrives at location. In this case we ask the hiking experience level and use external data, in this Komoot, to give a good suggestion for a beginners hiking trail. The use of such data and other sources is possible as we of course connect the chatbot to a data management platform in which all data sources from the region are accessible. We also give some hotel suggestions, because the user asked for it.

All the information captured in this phase is stored and can be used to personalize the following conversations even more.


At location:

When the visitor interacts with Alpi again at location, we already know that he/she is into hiking and the experience level. So if the person is ready to go a hiking trip, Alpi can give the most relevant hiking trail suggestion. Alpi also acts as the ‘hotel reception in your backpocket’. You can for example ask Alpi to look up a good restaurant and make the reservation for you. Again, this is possible because of the connections to the different data sources in the region from hotels, restaurants and the tourist office for example.


Evaluation phase:

Last but not least, you can also interact with Alpi when you’ve come home from your holiday. The objective here is nurturing; making sure someone would want to revisit and/or recommend the region to other people. Therefore, receiving feedback and reviews are also very important to assess the willingness if someone will share their experience in Engadin with family and friends.


Hence; Alpi can be the perfect, personal online assistant before, during and after your stay in Engadin.

All in all it was a great experience to be among enthusiastic startups, hackers and disruptors to see how we can innovate a traditional sector, tourism, in a traditional region, Engadin. With lot of new inspiration and energy we submerge ourselves into the new chatbot adventure!

Want to know more about chatbots, our vision on it and what we do with them? Contact us!

Share this story Linkedin Twitter Facebook