Smart cargo insurance — Solution for the Odyssey hackathon 2019

⁉️ The problem

Dumb containers

Only recently an accident occurred on a cargo ship near the coast of The Netherlands. A total of 270 cargo containers were thrown overboard into the Sea. A lot of questions have risen since then related to the traceability of these containers. After the accident, no-one really knew which containers fell into the ocean, what was in them and definitely not their location. These accidents still occur regularly. A similar story is the one about Porsche. They had to restart the production of 911 GT2 RS because the ship carrying 4 of them sunk.

Flat rates

Currently most cargo are insured against a flat rate depending on the weight of the cargo. They don’t even account for the value. If you’re shipping a container full of sand vs a container full of phones, if they weigh the same, they are insured for the same amount.

👀 Our solution

The shift

We see a shift from cost management to risk management. By focussing on reducing accidents and related damage, we can save costs and avoid unnecessary delays for both the insurer and the insuree. We can achieve this by providing risk-mitigating advice to the insuree. Together with enriching data like adding near-real-time ship tracking using AIS data from Spire, ( and later other providers ) we can more accurately calculate risk scores and give better advice.

Smart containers

By registering hand-overs, we can bring transparency to the current holder of goods between different carriers. Adding smart sensors to these dumb container will allow to easily capture detailed information about the goods across the hand-overs? All sensor data will be verified using a blockchain which will make sure these carriers can be held liable if the temperature or other sensor data exceeds the boundaries for too long. Combined with the other enriched data, all the parties on the platform get an accurate and trustworthy overview of the status of a certain transport.

Dynamic premium

Flat rates don’t work for all types of cargo. This is why we think that with all the data we receive from different sources ( eg. Spire, weather, container sensors, the characteristics of the goods, characteristics of the transport), we can more accurately create a risk profile. Based on this more accurate risk profile, better and more personalized advice can be given to the insuree. This advice can go from obligated to strongly advised to optional. Based on the type of advice and whether or not the insuree follows this, premiums can be adjusted as an incentive. This extensive profile allows for insurees who frequently get good risk profiles to get lower premiums because they are reliable.

Liability & faster claim resolution

By using blockchain to verify captured data and current holder of the cargo, we can very quickly see which party is reliable when something happens ( eg. Who had the container with bananas when the temperatures exceeded the boundaries for 1 hour. ). Since this data is trustworthy, when claims are being made, claims can be received and processed quicker and with fewer disputes. The quicker a claim gets created, the quicker other parties in the chain can adapt. This creates value for them because they immediately know if a certain transport will not be passing by them and give them the opportunity to find other transports instead.

Architecture

Our solution was pretty simple. We only had 2 developers in our team, so there wasn’t a lot we could do in 2 days since we also had to come up with something to build within this timeframe.

Our stack

TVM the cargo insurance company, who sponsored and led our track during this hackathon, already has a consortium of partners. Given that and the idea that we want to build an open platform where competition might join in the future and that all this data should be protected and only be read by specific parties.

We chose to use Hyperledger Fabric as our blockchain technology of choice. Our frontend consisted out of React and our backend out of Nestjs (enterprise nodejs framework).

The Application

transporter view

A transporter would be able to see detailed information about the status of its goods. They will also see the exact position and estimated arrival date using the APIs that were provided to us by Spire Maritime. Using their AI, we could also to a certain extent predict the future position of the vessel.

Besides this, they will also be able to monitor the sensor data of the container. Here, this will be the temperature measurements, tilt, the electronic seal and shock detection, but this can be extended with other sensors.

On the top right, the transporters will see their dynamic premium. Advice tailored to them with an incentive to lower the risk, and thus a receive a lower premium or the other way around when ignoring this advice.

An insurer will be able to see a similar view. But instead of seeing advice, they will see the risk analysis generated by the data fed into the platform. Based on this risk analysis, they can give better advice on risk-mitigating actions.

Challenges

Open platforms and especially platforms where competitors might join are hard to set up. Maersk is struggling with this as we speak. We don’t want to be another Maersk-IBM ( Tradelens ) tracking containers. Our focus lies on risk management, faster claim resolution and maybe even faster hand-overs by allowing for insurance to be taken over from a previous carrier. Tracking the containers are only a small part of the platform since they serve as a way to identify the carrier currently liable in the event of a claim. Tradelens could even be a part in our solution to capture even more information on the specifics of a certain container and their contents. They mainly capture data of what is happening with the container at the port.

We received a very good question at the hackathon: What does our platform have that makes sure we don’t have the same struggles as Maersk? — We believe that the incentive will be big enough for other competitors to join since they can potentially save costs by mitigating risk and provide a better service by giving advice. Claim resolution times will also be greatly reduced, which saves everyone in the processing time.

🛰 How did we use Spire maritime’s data?

We used the Spire Sense Cloud Vessel API in combination with Predict AI, their machine-learning feature capable of predicting positions up to 8 hours in the future. For the demo, we filtered the vessels down to those sailing under the Dutch flag for demo purposes. These ships get linked to transports that we have in our application. We can accurately position any of the ships we are tracking and sometimes even predict their future position up to 8 hours. This data can be used in our calculations for risk management. We can combine this data with other data like weather, in fact Predict AI already integrates weather into its predictive analytics. Using this, if we see a storm forming, we can give advice to the ship to take a different route or to wait in port to prevent damage or loss of cargo. Another example could be to avoid peak hours at the destination port because we know where ships are and where they will be.

We also received access to Spire’s Enhanced Vessel Data. Since we didn’t plan out a large user flow to fit our timeframe, we didn’t get to use this. But this database of technical details contains things such as specific dimensions and cargo capacity but also engine manufacturers,….. Our platform could later use this data to better calculate risk profiles for vessels since it will give us better insight into the value of the cargo and which vessels are more prone to delays due to engine failures or other conditions.

If you would like to try the API yourself, you can find it here or you can contact them if you have specific questions related to their AP

Every Dutch ship in The Netherlands (Spire Sense Cloud Vessels API & MapBox)