It could have been an April Fool’s joke but no, it was real. On April 1-3 Cisco and BeMyApp organized a hackathon where participants were given access to brand new Cisco technologies. Teams were given only two criteria to create their prototype. It had to include some sort of Cisco technology and respect the general theme of the weekend: "build the city of tomorrow!".
Five of our experts in IoT and AI couldn’t resist the call of a little challenge. And this hackathon was a perfect opportunity to try out our new technology. That’s how we ended up creating the "craftour" team with the help of Elie who we met at the event and who was our team’s expert in UI/UX design.
And so we spent a whole weekend in a building full of great developers using cutting-edge technologies to imagine solutions for the future. No flying cars but a lot of caffeine, crazy ideas, lines of codes and very bad puns: The Cisco hackathon was one of high level and one to remember.
Here at craft ai we all live in the wonderful city of Paris, full of history and monuments to see. You can literally walk round a street corner and find something interesting and new at any time. The issue is that you don’t always know where to look, or what’s that cool building you walk by every day. That’s where craftour comes into play.
We noticed that traditional guidebooks lacked flexibility. Indeed, in a guide you’ll find the generic stuff, but none of all the underrated content a city can offer. To find more detailed tips you can turn to community-created blogs and specialized content, but you still have to find the content that suits you in a sea of information. Plus none of those media help you during the actual visit, you don’t want to keep an eye on your phone or guidebook every two minutes to check if there is a monument around. To solve that issue you can use apps like Field Trip,that push you notifications with information on your surrounding to your phone. Unfortunately you’re still overwhelmed by the quantity of the content pushed to you.
As a result we came up with the idea of providing personalized content for tourists. You don’t want to push too much information. That’s why you need to filter it. It’s easy to record users’ preferences but you don’t want to base your recommendations on solely a user's profile, context also matters. Where is the user? Is it raining? How long has he been visiting? etc. Those are all very important parameters when pushing an information to a user. Our idea was to learn from the user what kind of content was relevant to him and in which context, to suggest only places and information that are relevant at the right time. To retrieve the location of the user we rely on beacons to be placed at points of interest. We also use the user’s location from his GPS if needed, but beacons are more precise and have the advantages of not draining the battery. Another advantage of beacons is allowing its owner to add custom content to describe the place where it is, and he can turn them off e.g. when an attraction is closed. To make the experience as seamless as possible, we use audio calls so the user can listen to his phone directly instead of having to read something and reduce the size of the app.
As for our business model, we intended to go for a B2B plan where businesses and administrations can improve their visibility by increasing their chance of being pushed. Be it a city’s heritage or a restaurant, because we push recommendation based on a user's profile they will have a higher rate of customer retention.
Screenshots of our final prototype; left is the home page where you can activate the app and call notifications; right is the page that is pushed when a user walks by a beacon;
The slides of our final pitch are available here (in French):
A little diagram of how the app works :
When the user is in range of a beacon, a notification is pushed if it matches his profile and local context.
For this hackathon, we chose to use the following technologies :
craft ai is a central part of the the project to learn the user’s habits and preferences. We used our upcoming learning API. It is still in closed beta but if you’d like to try it out don’t hesitate to shoot us a message.
Wezzoo is a crowdsourced weather platform. We use it to retrieve the weather at given a position.
DBpedia is a crowd-sourced community effort to extract structured information from Wikipedia and make this information available on the Web. We use it as information content for our app, in particular for monuments.
Connecthings provides contactless technologies in the public space, turning passive urban physical assets (street furniture, bus stops, public lighting…) into connected objects able to interact with mobile users. We used Connecthings beacons: We place them on the location users are likely to visit. Once a user stays in range of a beacon for a moment and it fits his profile we push him a notification on that location.
Tropo makes it simple to connect your code to the phone network with both voice and messaging. This lets Craftour call you and send you messages to enhance your visit experience.
We were offered a server on the Outscale cloud for the duration of the hackathon, so we put our backend there.
Unfortunately we didn’t finish in the top three spots, our dreams of holding a huge cardboard check were crushed... for now!
The 3 projects on the podium were:
- Gaia who created a gamified app that gives you reward points when doing a green activity in the city. Points you can then trade for coupons and other rewards.
- Eden for reimagining car sharing with cisco technologies and blockchains.
- And finally the winner was Smart Numa, who pitched a full solution to help trigger and operate emergency protocols during a fire outbreak.
Yet we didn’t leave empty handed. From the start of the event there was a joke challenge and I’m happy to announced that we won. It’s always good to see that even when working so hard everyone still has the time to crack a joke, reminding everyone that hackathons are for fun after all. As a prize we got an AwoX striimLIGHT color that joined our showroom.
The winner of the #savoirflare challenge
Even though we didn’t win we are satisfied with what we accomplished in such a short timespan. We had lots of fun, learned about some cool tech, and we believe we had a product idea that could work. We’re also happy that we found another use case for the craft ai technology and that it meshes so well with others. We may continue developing our prototype as it’s a great showcase of what craft ai can to do
For the hackathon itself a lot of projects made good use cases for IoT and from what we see most project could benefit from AI.
The future seems full good of surprises for IoT and AI, we can’t wait!