We met over Slack and originally planned to work on a completely different project, but after discussions and brainstorming, we felt inspired by the Swisscom API and started planning an application around the data. During the hackathon, we found the Huawei mobile services, and the final plan for our app was born.
The fantastic team behind the app is:
Inspiration and motivation
We realized that during the COVID-19 pandemic, some new challenges have emerged that block restaurant businesses. It also became necessary to consider security and safety standards while providing customers with good suggestions for places they can go that meet their preferences. So, we wanted to build:
Crowdfree - a recommender system with a fabulous UI that can recommend great restaurants and cafes that are free from crowds.
The Food For Thought challenge by IBM and SwissRe felt inspiring as our app would be about food and restaurant services. The Helsana challenge about enhancing digital services with data is also close to our project as we can use their data to create more targeted services for our customers.
What the app does
Crowdfree is a mobile app for enhancing the restaurant experience during the pandemic. We have built a Multi-directional Recommender System. Most recommender systems only consider the user's preferences while giving suggestions, but we focus on how crowded the restaurant is as well. With this solution, we aim to offer the best and safest user experience.
Crowdfree offers a fully automated service to give the best user experience when visiting a restaurant or cafe. The user starts by looking for a suitable seat to book, goes there, and checks in to get the online menu. After ordering, the user pays and checks out through the app. Finally, the user fills out a survey to improve their suggestions and share good venues with other users.
Customer Segments: Crowdfree targets tech-friendly people aged between 25 - 50 that have internet access and a mobile device.
Value Proposition: We offer a quick, easy, and customized way to finding less crowded restaurants. For this, we develop a multi-directional recommender system that considers both the user preferences and how crowded a place is. A recommended item would not be frequently recommended to other users in order to avoid crowding. The recommendations are customized based on user preferences for restaurants, gamified, and can offer a chance to win points and vouchers as a motivation to use the app. If we can convince users to use the app to check-in and check out we get more accurate information about how crowded a venue is, which benefits all users.
Channels: We can reach the users via our mobile application. Also, we plan to advertise on social media platforms, e.g. Instagram or Facebook. For offline advertisements, we can use billboards, outdoor advertisements, and flyers to establish a strong social presence.
Customer Relationships: We offer the users an online restaurant booking platform, an online menu with ordering, a recommender system, a check-in/check-out system, and online surveys.
- Vendors - restaurants and cafes
- Retail shops and entertainment venues (they can offer vouchers in-app)
- Swisscom - Mobility Insights for predicting crowds and improving our data
- Azure - Backend hosted on Azure Cloud Services
- Huawei Mobile Services - enhancing the customer journey with HMS Core APIs
Revenue Streams: We are a B2B business and businesses can sponsor their venue via our application to reach more clients. We also have a pricing plan to suit all our customers.
PLAN A: 1400 € / month + 3% (A percentage of sales) PLAN B: 1000 € / month + 5% (A percentage of sales) PLAN C: 700 € / month + 8% (A percentage of sales) PLAN D: 400 € / month + 13% (A percentage of sales)
How we built it
We have built a restaurant booking app. At first, we look for nearby pleases based on HMS Location and map kits. Then, we developed a system (check-in/ check-out) to know the approximate number of guests per restaurant. Based on this system and the API from Swisscom we guide our users to the most suitable place based on their preferences.
When the user reaches the place, we ask him/her to scan a QR code that was built with the HMS scan API. This QR code is attached to the check-in system and the online menu. Moreover, he/ she pays via our payment getaway and check-out. Finally, we ask him/ her to fill out a survey to improve our recommender system.
Live app debugging at Hackzurich with Huawei Mate 30 pro
The recommender system is combining collaborative filtering and content-based filtering for a hybrid system. For the content-based part we used Zomato Restaurants Data and we built the system with python and RESTful API to integrate it with the mobile app. On the other hand, collaborative filtering will be developed after getting users during beta testing.
This is the user experience Demo Video
Challenges we ran into
One of the problems that we faced is to know the number of guests along the day for each restaurant. It was challenging especially since Google Places API doesn't support an estimate of how long people typically spend at a restaurant or venue.
We are a team of 5 and 4 of us are participating virtually. To overcome this challenge we created a well-rounded workflow and divided tasks between us. We are also keeping regular meetings and collaborating over videocalls.
How to create a simple user experience with all these challenges and the tech complexity.
Accomplishments that we are proud of
We are proud that we have participated in this competition competing with people from all over the world. Furthermore, this Hackathon helped with meeting other incredibly talented people like us, working as a team, and taking on challenges that put our problem-solving skills to the test.
We have successfully created a full prototype during the Hackathon. Moreover, our system has been deployed as a real-life proof of concept, and it could be used easily. Furthermore, we made a business model, so that we can commercialize our app.
What we learned
By joining the workshops, we learned a lot about services provided by sponsors like Swisscom, Huawei, IBM, and Helsana. For example, from the Swisscom API we learned about the Swiss special coordinate system! Huawei helped us out by borrowing a fantastic Huawei Mate device for us to test the app on! The Microsoft team sent our team very useful documentation that helped us while developing the MVP. We learned how to use the Microsoft Azure platform to deploy our application. On the academic side we read some really useful articles about recommender systems and the necessity of improving them.
Backend deployment on Azure:
What's next for Crowdfree
- Establish real customer relationships with restaurants and cafes.
- Integrating more data from the Swisscom, IBM/SwissRe and Helsana data and APIs in order to improve our recommendations.
- Continued development of the mobile app and backend.
- Getting in touch with angel investors and plan the next steps for achieving a product that can be released.