The last 2 days I attended Hackbay 2019, which is an exciting hackathon organized by Zollhof. But what exactly is a hackathon? In this case, big German companies (Siemens, Schaeffler, HUK Coburg, Nürnberger Versicherung, VAG) presented business challenges they are currently struggling with. Small hacker teams of 4-8 people chose their challenge, thought about a creative and innovative solution, and then tried to wrap it in a business model and working prototype which, at the end of the hackathon, was presented as a pitch in front of a jury. This all happened in 2 short (or very long?) days.
Hackbay offered 5 challenges:
- Damage Claim made easy – Data Discovery Track, by HUK Coburg
- Let’s shape the future of mobility – Data Discovery Track, by VAG
- Get on the bike of the future – Augmented Logistics Track, by Schaeffler
- Activity-Based-Working is the new thing – Smart Workspace Track, by Nürnberger Versicherung
- Shape the workspace in industrial production – Smart Workspace Track, by Siemens
At the start of Hackbay, these challenges were presented, and afterward, all hackers could choose their favorite challenge. I decided to participate in HUK Coburgs challenge, because their pain point was very easy to understand (current damage claim process for car owners is very inefficient, takes a long time and a lot of manpower + the insurance struggles with extreme weather, such as hail, wherein lots of customers are trying to contact them).
All hackers that were interested in HUKs challenge gathered in one room (about 30 people) and brainstormed ideas for tackling this challenge. Once the first ideas were exchanged, people that believed in similar concepts started forming teams. This is how our team of 8 people slowly evolved. Then it took another 2 hours until we were all on the same page and knew what we would like to achieve within these 2 days.
Many other teams tried to focus on making the damage claim process easier by using chatbots or app ideas that would guide you through the damage claim process. But we wanted to completely automate this whole process and make it as easy as possible for the insurance and its customers. This is how the idea of the Claimbox (also our team name) was born.
Claimbox is a small device with a variety of sensors (acceleration, vibration, microphone, gps, …) which gets attached to your car and is powered by a battery. It always runs, even when the car is parked. And it was exactly this which was our main use case: The car is parked, it storms outside, hail is falling down and your car gets damaged by hail.
Usually, this is a horrorific scenario for every car insurance company and their customers. The insurance has hundreds of people trying to call them and claim their damage. In addition to that, they have to start organizing appraisers because for such a big scale event it isn’t easy to get individual appointments with appraisers. Right now this process is very slow and the customers are very annoyed because they don’t know what to do with their car and they can’t reach their insurance company because many other customers are trying to call them after the hail damage. They grow impatient and mad at the insurance because they are paying lots of money monthly, yet when they are finally in need, they experience bad service.
Now with Claimbox, every customer has a device in their car which automatically detects if the car got damaged. Once damage is detected, Claimbox sends this information to the insurance company and the insurance company can instantly act on it and proactively contact the damaged customers with a solution for their problem. This is a classic win-win. The insurance doesn’t have to ramp up their service center because way fewer people will be calling due to the automatic contact they receive, instructing them on how to proceed with the situation. This way, the insurance company can save a lot of money, meaning the customers have to pay less per month. So on the customer side, people are happy because of the lower fees and because of the excellent service (no more hour-long waits on the phone; now one can be certain that the damage gets taken care of).
How did we implement our prototype?
We used a Raspberry Pi and attached different sensors. A vibration sensor and a microphone (at least in theory – unfortunately, we didn’t have a microphone that could connect to the Pi). For each of these sensors, we coded algorithms for the hail detection. The vibration sensor works with a simple threshold implementation and the microphone sound outputs are analyzed by a Neuronal Network, which we trained on YouTube hail sound data and Kaggles Urban Sound Classification dataset.
If both Sensors detect a certain damage type (in our case hail, but we can also imagine many other damage detections, e.g. minor collisions when your car is parked), the data is sent to a Flask-based web server. The server receives the cars insurance number, the damage type, the cars location, and the timestamp. It uses this information to double check with weather services about whether or not it actually hailed in this area. If it did, the insurance company is informed about the hail damage immediately. This process takes about 5 seconds – from the hail damage happening, to the insurance company knowing about all damaged customers. Currently, insurance companies need about 2-3 days to estimate how big the hail damage is in a certain area, but with Claimbox they can do it within minutes.
This process also triggers an automatic notification to the customer. The customer gets informed about the damage, receiving that everything is taken care of by the insurance, making it so that he never has to worry about what to do with his damaged car.
At the end of Hackbay, each team had 5 minutes to pitch their idea and 3 minutes for a Q&A session. As mentioned earlier, although other teams that participated in this challenge worked on chatbots or innovative app ideas, we were able to convince the jury with our idea and won the HUK Coburg challenge. Great team!
All in all, it was a super interesting event where you could meet many like-minded people, work on stuff you are interested in, and create something that could actually improve a current, real-word situation. Great job, @Zollhof, for organizing this event. It was a challenging event within a great atmosphere, big recommendation for everyone who is interested in tech and innovation.
I help companies with data analysis, business intelligence, machine learning and many other data related fields.
- Bundestagswahl 2021 – Twitter Analyse
- Tableau Dashboard: How to get your Vitamins in Winter
- My Interactive CV
- Where are Germans Flying?
- Domestic Flights in Germany – Facts Overview