Inspiration

Enhanced by new anti-COVID regulation, organizing a trip has become increasingly complex. To understand the complexity of what you do as traveler every time you go for a trip, just imagine how many apps, website, notifications and services you might use in a trip from your room in Dubai to your Airbnb flat in London (Ticket, Check-in, Taxi, UBERs, FastTrack, COVID test, flight updates...). You can't count them, right?

Well, we want to make such experience as smooth as possible and for this reason we decided to build a non-app enabled by Artificial Intelligence.

What it does

You might ask "what do you mean with non-app?". Well, a non-app is essentially an app which requires no interaction. We don't want our users to search again and again for our app in their over-app-populated phones. We rather aim to embed our service in the tools they use on a daily basis, such as Whatsapp, Facebook, Email, SMS, Slack etc.

This led us to ideating a chatbot which will assist any traveler along their trip. Rydaround will reserve you a cab at the arrival airport or buy a train ticket simply by chatting with you, sending you reminders or simply provide that essential information you need any time you are going for a trip (such as COVID policies or weather forecasts).

How we built it

In order to achieve a non app, we have to minimize the in-app interaction with our users. Yet, we want to be close to them and have a personal relationship. Therefore, in the moment they subscribe to Rydaround, they will receive a Vcard with our contact (email and phone number) to be saved in their phone.

When they buy a ticket from any airline, the only thing they will have to do to initiate our service is forwarding us the confirmation email. From the ticket we can extract all the information needed to take care of them. A few days before the flight we will remind them to do the check-in, we will remind them to do a COVID test if required or we will suggest them to take some warm clothes if it is forecasted to be cold at their destination; a few hours before the flight we will check the estimated time-to-gate and we will ensure they have arranged their transfer to/from the airport; we will check if they are on time and provide fast-track services or if their flight has been delayed so they can relax. At their destination, we can recommend activities, restaurant and amenities. And it all starts with a message on

All the service have been build on a Django/Python server. The Django/Python server has two main functions: 1) querying the APIs of our partners (such as SITA's APIs) to collect the information and 2) push to the user the service required. An example of a service would be "Hi there, I understand you want me to book you a taxi to go to the airport? Please, reply YES to confirm or NO to cancel." or "Hi there, I have checked the weather forecast in Amsterdam and it seems it is going to be quite cold. Bring some warm clothes with you."

Between the server and the user there is an Artificial Intelligence layer currently developed with DialogFlow. The role of the AI is: 1) engage with the user inbounds request using NLP and create a human like conversation; 2) collect information to identify and activate the service the user is requesting. So, if the user asks "At what time shall I be at the airport?", the AI engages in a conversation to successfully arrange the transfer to the airport preferred by the user.

Challenges we ran into

We mapped out most of our processes in detail and we were able to prototype the solution quite efficiently, however we realized that using DialogFlow might not be sustainable from a business model perspective and hence developing our own NLP should be the way forward. We explored such route but it would not have been possible to develop it within the time constraints of the Hackathon.

Accomplishments that we're proud of

Being able to develop a working prototype in such a short amount of time was definitely a challenge, given that we could not dedicate ourselves full-time due to other professional commitments.

What we learned

The refined the architecture of the solution NLP+Django Server during the project as we realized that the original concept was not feasible to manage inbound/outbound context in efficient way.

What's next for Rydaround

Improve the services currently built as prototype and identify new ones. Define go-to-market strategy and refine business model.

Built With

Share this project:

Updates