Insurance claims just got smarter thanks to AI — BlockClaim promises unparalleled fraud detection

Antonio Agudo
Published in
9 min readMay 14, 2018


It is widely believed that the first known case of insurance fraud occurred as long ago as 350 B.C., starring a man named Hegestratos. This greek merchant took out an insurance on his cargo and ship and subsequently planned to sink it intentionally to claim damages.

We only know of him, because he was not very successful. When his passengers learned that he was going to sink the vessel and drown them along with it, he jumped ship during the incipient mutiny and drowned right then and there.

Since then, insurance fraud has evolved into an elaborate kind of cat and mouse game, developing alongside the status quo of technology. New and clever methods of defrauding insurance providers were devised and insurances in turn came up with increasingly complex forensic approaches to investigate the elaborate schemes of suspected fraudsters.

Now however, it seems that insurance providers could finally gain an edge over fraudulent behaviour thanks to the advent of artificial intelligence and blockchain technology. Incumbent IT solutions are complex though and threaten to impede or slow down the adoption of these technologies considerably. This is where BlockClaim, sees its unique selling point — helping large scale systems integrate to this new kind of approach to unparalleled levels of claims automation and fraud detection.

Interview with Niels Thone

Niels Thone co-founded the company BlockClaim and took the time to discuss how they plan to shake up the insurance claims processing space.

Niels Thone, Co-Founder of BlockClaim

W: Niels, what is the background of your team and where did the project originate?

Niels: We are actually a spin-off from Imperial College, London. One of our team members graduated from Imperial College and did his dissertation on blockchain telematics and insurance. He developed a proof of concept application in that phase and since he is in a group of friends of mine he talked to us about it and wanted to have our thoughts on the business side of things.

So, after a rigorous due diligence phase, where we analysed both the business side and the technology, we decided to pursue this opportunity. In this phase, our CTO Greg, who has lots of experience with software development, came on board. The guiding Professor at Imperial College found our idea intriguing and started to introduce us to potential investors and insurance people.

W: Tell us a little about why you chose to work on this particular project? What is the idea behind it?

Niels: The basic idea behind it is simply helping people. My family is heavily engaged in positive entrepreneurship in Belgium so those values of ‘helping the customer’ are very dear to me. If you’ve ever had an incident on your car, you know how time-consuming it can be to get your money back. I believe we can reduce the kind of hassle you have in that space. The average time it takes for a claim to settle is about 25 days in the U.K. and it’s not a lot different in the rest of the EU. I believe we can reduce it to less than a week.

And obviously there is the B2B aspect of it where you see an opportunity and go: “Wow, so this is costing insurance billions per year”. Over 10% of yearly revenue goes to the problems of fraud and claims automation. As an entrepreneur, I knew we could make an impact and just went for it.

So, on one hand it’s really an exciting technological issue where there are many parties involved and fraud in a way is organized crime, where actors get smarter and smarter. So, it is intellectually stimulating to do this type of work and you get to kind of fight crime.

On the other hand, the commercial person in me sees an opportunity and says “Wow, we can really do something here, we can build a beautiful business while we help other people.”

W: What is wrong with the current approach of doing claims?

Niels: There is several things that are actually wrong apart from the fraud.

You can divide this into two categories — one is the category of problems of the customer, the other is the challenges of the insurer.

To start with the customer — if you make a claim, it’s often really low-tech and backwards. You have to phone the insurance company, send them an email then a letter and whatnot. The way to commence the claim is already not easy by default, since it includes a lot of information. But when the communication channel also has slow processes everywhere and it’s just not a nice customer experience, it gets frustrating very fast.

But it is even worse for the Insurers. In a claim there are typically multiple parties involved and their data is coming in from very different sources that are often not matching in terms of data. That means you can’t combine them in a way that could be automated at the moment, which is inefficient.

To make matters worse, all those parties have to give some input in this process. There is the counterparty insurance, a fraud expert, a lawyer, potentially a garage that has to fix the car and a car rental company that will give you a replacement car. Police and ambulances might also be involved. In the current process, you have to wait until the last party completes their task for the next party to move on, so it is very sequential. That also means individual participants get quite a bit of power over how long the claims process actually takes, which is what can make waiting times longer, because it’s really inefficient.

The final issue is, that no matter if you drive into a pole and dent your bumper a little bit, or if you have a massive accident, all of the supporting documents of the claims case go through one point of contact. This is for the most part one single person who has to sort through all of that unfiltered volume of data with not much automated help from IT systems. This puts an upper limit on how efficient such a claim systems can be.

W: How do you achieve better fraud protection and claims automation than what is currently possible?

Niels: One of the most exciting opportunities of fraud detection and claims automation is the rise of telematics. In the last five years it has become a big thing in the U.S., but now all the major insurers do it. Basically, you have a black box in your car, or an app on your mobile phone that you can use in your car, which enables a new kind of business model called UBI (usage-based insurance).

It means you pay for how you drive: You drive safe, you get a discount. You drive recklessly, you pay more.

Especially millennials are fond of this business model. It means that you measure how you are driving. So, for the first time ever, there is an objective data source about driving behaviour. That means if you have an incident, you have data around the incident like how fast you were driving, were you accelerating, what was your relative speed et cetera. So, for the first time ever you don’t have just situational proof, where you rely on witness testimony and they can lie to you, you actually check data around the incident.

We use an artificial intelligence system that detects fraud from telematics and claims data and by using this objective data you can create dynamic rules and scores for claims. So, in essence we calculate a risk profile to augment all this data and use it as a stepping stone to automate the claims process.

After all, if you have this detailed risk score you can, for example, put all of the lower risk ones into one bucket and say, “We don’t have to run this by our fraud experts anymore, we might as well just pay these customers out immediately.”

This not only reduces the payout time for the majority of customers, but since you have this set of rules, you can send the claims to the right people within the organisation, which further reduces overhead.

W: How does blockchain come into that??

Niels: The blockchain serves a few purposes. After calculating the risk profile through our AI, we feed this into the blockchain so it can be audited.
Also, we use the blockchain to settle the claims in a quick and efficient way. We use it as a single source of truth. In a regular workflow, you have to wait for emails from each and every participant to the claims process to come in and then check all your boxes if everything is in order. If you have this single source of truth, then people can work on it simultaneously. And if you connect that to a notification system, people (including customers) who have to be in the know, can be kept informed about the progress of a claim every step of the way if they so wish — which is a much better UX or customer friendlier approach.

Also, right now, the way the process works, with all the parties involved in a claim, everything is based on trust. The individual steps are not transparent. For example, one of the insurers we are partnering with noticed that, at times a garage where a car from a client is being repaired for three days, gave a rental car to the client for six days. The reason being, is that there can also be internal fraud within the value chain of the claims process and the car hire company figured out a way to make some extra money. The only reason the insurer got wind of this, is because they talked to customers. But in a blockchain system everything becomes an auditable data point, so it is much easier to detect this kind of fraud.

W: Why have you picked automobile insurance claims as your target market?

Niels: Obviously there are many insurance markets where this could be applied to. Life insurance, home insurance etc. would also be a good fit. But fraud rates are highest in the car insurance business, so that is why we chose to start there. We very much intend to expand because we know our engine can be applied to other processes. We were approached by some very large corporates recently and they asked if we could solve some of their problems that are related but not exactly on our roadmap right now. Since we are a startup we only have so many resources though and we do want to focus on “doing one thing only and doing it well”.

W: What is the impact you hope BlockClaim will have in this space?

Niels: There is a potential of having a massive impact. In terms of fraud and automation, if 100% of the market would hypothetically use our solution — obviously that is almost not possible, but for the sake of the argument let’s say this is the case — there would be an upside of 1.5 billion British pounds that is lost to fraud each year in the UK that could be reclaimed. So if we are able to eliminate crash for cash fraud and all these types that we are really focusing on, premiums of insurance clients could be reduced by 90 pounds each per year. Lowering premiums on such a scale is a massive competitive advantage for every insurer in the market.

W: Where is your product roadmap taking you?

Niels: We have developed the product and are testing it in the market with several partners in Europe and probably will be working with a partner in the USA as well. The goal is to get a refined and scalable alpha version out by the end of the year.

The underlying vision though is about self-driving vehicles and an evolving transportation market. How is the claims process going to look like in the next 5, 10 or 15 years. We plan to be in the center of this paradigm shift.