Matmut
Quantum exploration to begin in 2023
When we first contacted Matmut, known for its culture of innovation, the idea of exploring the potential of quantum computing was not yet in their plans. They were already working on several data-related use cases, exploiting their vast volumes of information to refine their processes and make informed decisions. However, certain limitations of traditional technologies were holding back their ambitions. It was when we told them that quantum computing was already available that the technology really caught their attention.
The meeting: brainstorming and exploring a new field of possibilities
During an initial brainstorming meeting with Matmut teams, we presented the various quantum technologies and their specific advantages. This first fruitful exchange opened the discussion on potential use cases that quantum computing could unlock. Together, we identified several areas where this technology could bring real added value, constituting a selection of use cases to be explored in greater detail.
From concept to action: use case selection
Among the various possibilities identified, one of the most promising concerned motor insurance pricing. Traditional methods, while effective, were reaching their limits when it came to establishing rates, based on vehicle risk classes, that were both fair and explainable. Since quantum computing could be a solution capable of overcoming these limitations, we decided, in collaboration with the Matmut team, to focus our efforts on this specific challenge. The aim was to assess the ability of the quantum approach to improve the accuracy of vehicle classifications, but also to make the process more transparent and understandable.
The goal: to create business value
In our approach, it was essential to define key performance indicators (KPIs) that would reflect not only technical gains, but above all the real business value. This is an essential exercise, especially in the context of an innovation project, where results are sometimes difficult to quantify! This collaborative effort enabled us to identify relevant KPIs, such as improved traceability and transparency of algorithms, enabling us to structure the project around clear, measurable objectives.
Exploration: Proof of Concept(POC)
The POC was then launched to test QbitSoft's1st quantum module "Q-MEDOID", developed by our R&D teams. This is a "quantum classification" module which, in this use case, enables vehicles to be grouped into different risk categories according to different criteria and characteristics. We worked with a dataset supplied by Matmut, and conducted extensive tests. Each group of vehicles, represented by a real vehicle, enabled us to simplify the classification and then refine the proposed rates. The final deliverable took the form of a detailed report, explaining our approach, the mathematical tools used, and the test results. The results were sufficiently convincing to enable Matmut to validate the potential contribution of quantum computing to optimizing its automobile pricing.
Impact: towards large-scale transformation
On the strength of the success of this first stage, Matmut has decided to continue the experiment on a larger scale, applying the algorithm to all the year's actual data. The aim is to validate the economic gain hypotheses, thus solidifying the pricing strategy, and to move on to the implementation phase with the tools already in place at Matmut.
In parallel, a second use case has been launched: the detection of fraudulent activity patterns before reimbursement.
The adventure is just beginning!