Meet DreamQuark, a French start-up that wants to help banks, insurance companies and asset management companies for all their artificial intelligence needs. DreamQuark crunches your data, creates models based on machine learning and lets you apply these models to all past and future data points.
Many banks still use COBOL code to calculate all kinds of financial transactions. COBOL is a reliable programming language, but the only problem is that it was designed more than 50 years ago. Now that all industries are considering machine learning as the next big thing, banks are being left behind.
At the same time, banks and financial companies also generate a ton of data. These companies are already taking advantage of these data with traditional linear regression algorithms and simple criteria. And yet, you can dramatically improve these patterns by switching to machine learning.
DreamQuark has been working on a product called Brain so that financial companies can make smarter decisions. There are many cases of use of algorithms powered by artificial intelligence in the finance industry.
This is a great way to assess the risks, from fraud detection to the fight against money laundering and credit scoring. It can also help you manage a portfolio by detecting early signs of market changes. And DreamQuark can also segment your customer base and detect certain models. In this way, you can promote financial products and improve retention rates.
"Thanks to our global approach, we are faster than companies specializing in fraud in a particular country, for a particular chain," said co-founder and CEO Nicolas Meric. And it's true that a customer usually has several financial products – everything is connected. Your credit history could be a sign when it comes to anti-money laundering detection for example.
DreamQuark also ticks all the right boxes when it comes to compliance because it can identify bias and explain each decision to comply with the regulators and GDPR. And of course, DreamQuark does not share data between different clients. You can let DreamQuark manage the service for you or install it in your own data center. But everything is segmented between each customer and the data is protected.
The company just raised 3.5 million dollars (3 million euros) from CapHorn Invest and Plug & Play. Up until now, DreamQuark has attracted a dozen clients, including BNP Paribas and AG2R La Mondiale. And I am sure that every contract is worth a lot of money.
By focusing on financial services, the company can quickly deploy its solution and build models because many financial companies share the same needs.
But building models is just the first step. Financial companies can then use the DreamQuark API to evaluate all future data. For example, how many times a year do you have to call your bank because you made a suspicious purchase with your card, but that was actually you?
Startup can detect these false positives much more accurately so you do not get stuck on your own bank account. And this is not a security compromise because DreamQuark also identifies fraud more accurately.
On the competition side, I thought that the banks themselves would be a serious threat because they have a ton of money. But it seems that they have already tried and failed to set up teams of data scientists to work on these issues. "They compare how much it costs to build internal projects and their success," said Meric. "We provide services that you will not find in any open source framework."