Sites Inria

Version française


CP - 27/09/2019

From research to prototyping - let’s take a closer look at a partnership between a start-up and a project team

Startup Mieuxplacer Startup Mieuxplacer - © Robin Finance

With an academic background in applied mathematics, Matthieu Sénéchal co-founded the start-up Mieuxplacer three years ago, a platform which aims to democratise wealth management consulting through the development of recommendation algorithms. Although this young company, which employs around thirty or so people at its sites in Bordeaux and Paris, does carry out operational research, making technological developments on its own, on this occasion it sought assistance from the Inria Bordeaux - Sud-Ouest research centre. More specifically, it was the project team RealOpt, a joint undertaking involving the University of Bordeaux, Bordeaux INP and the CNRS, which assisted the company, helping to improve its existing tools and to offer users an effective solution.

The design brief was clear: to solve a combinatorial problem to match user profiles to types of financial products. Seeking an optimal response, the start-up turned towards François Clautiaux, head of the RealOpt project team. Together they were able to determine what improvements would need to be made to the existing algorithm. Using proposals from mathematical modelling, Rémi Duclos, an Inria engineer, developed three possible methods, testing each for durability. Duclos found this a real challenge from a technical perspective, given that it involved solvers and a language with which he had never worked before and a field that was equally alien to him: the financial sector. Beyond the necessary upskilling, it was the human factor that would prove most important in this collaboration:

 Facing us were experienced technicians with proven knowledge of the world of IT. It's brilliant getting to work with people who really know what they’re doing. We provided our expertise in combinatory modelling using a lexicographic approach, which had never been done before. In the financial sector it is rare to employ the use of integer optimisation. We feel as though we met their demands, delivering exactly what they wanted

explains Rémi Duclos.

The prototype and its analysis report were released after two months and incorporated into the platform. A success story for Matthieu Sénéchal, Mieuxplacer’s director of research and development:

 I am pleased with how everything turned out with this partnership and the speed we were able to work at. For a start-up, that’s a real advantage

It is true that being able to work quickly was one of the strengths of this collaboration. In January 2019, discussions began on the delivery of a functional prototype midway through the year.