10 years of Vekia: the success story of an Inria start-up

Publish on 22/01/2020
Just ten years ago, Manuel Davy, co-founder of the Sequel team from the Inria Lille - Nord Europe research centre, launched his start-up: Vekia At the time the small company, specialised in machine learning, proposed consulting services to the major retail groups in the Lille region. Today, Vekia is developing a supply chain solution for stock management. It employs approximately 50 people, works with around 20 major retail groups and, in 2017, raised 12 million Euros in order to expand.

What led you to create your business in 2008?

Logo Vekia 2018

At the time I was carrying out research in machine learning at the CNRS (French National Centre for Scientific Research) and at the Inria Lille centre with the Sequel* team. I very quickly wanted to get more involved in industry. We rapidly had contacts with retail companies, of which there are many in the region. Several companies showed an interest in the team's research, and we began to work with them on prediction tools. I then wanted to create a start-up, to develop this service further. And that is how Vekia was born. In the beginning, we provided consulting services to the IT departments of the major groups. We helped them with issues of fraud detection, personnel management and stock forecasts. This lasted two to three years, and then Vekia ended up focusing solely on the creation of stock management software.

What does the Vekia solution consist of today?

Stock management is the thorn in the side of retailers. The French retailers use old solutions, even though they have to face up to competition from giants such as Amazon, Cdiscount, Alibaba, etc. It is a tough challenge to take on. On the Internet, consumers want to receive products very quickly. For the seller, this implies having stock all over France and for a very large number of product references. Traditional solutions are incapable of managing this. We therefore propose giving an artificial intelligence system the complex task of managing stocks in a very accurate way, in each point of sale, each warehouse, for each product reference. Today we have the most advanced machine learning solutions in the world for stock management for the retail sector. We have also gradually opened up our market to after-sales companies, for the management of spare parts.


 We must cultivate our Start-up Nation spirit




Manuel Davy

In what way is your solution one of the most innovative in the world?

From a purely technological point of view, we are developing our product on the most advanced technologies, like Spark (distributed computing). Next, we are working on a very large volume and a great variety of data. They come either from the clients themselves, or from external sources, like the weather for example. We also have statistics collected by social network partners. We do not use personal data, but trend data such as the number of likes on a product, for example. Finally, we are adapting to all the latest uses of e-commerce, such as "click and collect".

What also gives us an advantage is that we are experts in machine learning. We are not mere users; we contribute to the development of the new algorithms.

Let's go back to Vekia's beginnings. How did the creation of the start-up go? How were you helped to get started?

At the start of the project, the Inria Lille centre financed an engineer position which helped us to do our first trials. It also provided us with an office and helped us to establish our first commercial contacts. I was also assisted by the CNRS, thanks to the 25.1 scheme which, for a certain period, allows you to create a business whilst remaining an employee of the organisation. I received a salary, to be reimbursed by the company at a later date.

Was it difficult to go from the world of research to that of business? 

I personally really enjoyed it. At the end of the day, being a researcher prepares you quite well for the role of company boss. You need to be factual, be at ease with figures and with abstract and complex concepts. Researchers also do marketing when they publish papers, or go and 'sell' their research in conferences... Then, commercial flair and management are things that you learn on the job, they come with experience. Personally, I have always been interested in business creation. Throughout my research career, I was in contact with the world of business. I have always been quite entrepreneurial in my career as a researcher. Ultimately, I think the key is not to be afraid of not succeeding.


 Being a researcher prepares you quite well for entrepreneurship



In France, artificial intelligence is currently the subject of a government action plan with, in particular, the Villani report. What is your view on this subject?

It is good news that the government is taking up the issue. The Villani report is an opportunity for France to take back technological and scientific leadership. Organisations like Inria or BpiFrance are extremely important for the success of the scheme. We must cultivate our "start-up nation" spirit.

There are, however, two very important points in my opinion. France must carry out very specialised artificial intelligence because, for very general AI, the place is taken by Microsoft, Google, etc. The other important issue is training. For me, this must start at nursery school, because the citizens of tomorrow must understand artificial intelligence and be capable of interacting with it, whilst knowing how to interact between humans even better.

Next, I am convinced that scientific higher education institutions must imperatively reinforce scientific training. If we want to be a start-up nation in the field of AI, we must train creators of AI and not users. And this requires a very high scientific level.

*Sequel is currently a joint project team with the CNRS, University of Lille - Sciences and Technologies and University of Lille − Human and Social Sciences.