In these recent times, the economy is undergoing through many turbulence. While scary, the economy must go on to prevent further collapse. To be able to keep going and not collapsing, businesses require insurance. Yet, traditional insurance relies heavily on banks, criminal records and personal financial data to determine risks, as a result, only a small percentage of people can get insurance.
So, how can we insure everyone without discrimination? Managing risks for insurance is a tricky business without solution. With the emergence of Blockchain technology and Machine Learning, new concepts are being tested to solve this problem. That's where Trench comes in!
What it does
Trench is a ML automated insurance Dapp where people even without banks can get insured. There are no private data required in order to get insured.
Unlike traditional insurance companies, Trench heavily relies on real data to determine market conditions. Instead of determining risks by guessing from someone's financial data, Trench uses market data to predict and calculate risks. As a result, Trench is able to insure everyone while having dynamic premium pricing based on market conditions to keep it's fund pool afloat.
How we built it
We collected thousands of historical inflation data which we then cleaned and composed to make a working ML model that can predict the outcome of inflation in the coming months.
After that, we worked on the smart contract and frontend. Here we spent a lot of time to fix bugs and solve uneccessary problems, but it was all worth it!
Challenges we ran into
The lack of documentation on JS sdk got our smart contract running around in circles. Fun fact, we didn't know that the js files needed to be built into WASM files. It took Axel 3 days to figure that out because uploading Typescript file as a smart contract didn't give any errors.
Accomplishments that we're proud of
We are proud of the results. We managed to get the proof of concept done even if it's in late submissions. Thank you Near teams for the generous late submission!
What we learned
We learned a lot about Near! While it is very hard at first due to the huge amount of differences with traditional blockchain networks such as Ethereum, it is pretty easy when you get the hang of it. We are looking to learn more and if possible, help update the JS docs :)
What's next for Trench Automated Insurance
We look forward to your feedbacks, if this project is interesting and feasible to you guys, we would love to bring it to the mainnet and bring this to be used to solve real world problems!