How can we bring generative AI closer to the needs of SMEs? This is the mission that Léo Benkel has set himself. Photo: Matic Zorman/Maison Moderne (archives)

How can we bring generative AI closer to the needs of SMEs? This is the mission that Léo Benkel has set himself. Photo: Matic Zorman/Maison Moderne (archives)

Four years after Pure Lambda, Leo Benkel has set up his second company, with Thomas Friederich: The Artificial Business aims to facilitate access to artificial intelligence for Luxembourg (and European) SMEs on a sovereign basis.

How can we both distance ourselves from the American players in artificial intelligence and make life easier for European small- and medium-sized enterprises that still want to automate a number of tasks? With The Artificial Business. Since the summer, has been embarking on the new project with the ulterior motive of "restoring sovereignty to European artificial intelligence".

How can he do this? By offering a platform that allows companies to "download" AI agents to their site or to their cloud provider. "We use all open source models, including those from Meta, Google, Amazon and Alibaba", explained the young entrepreneur, who just happens to have a degree in AI agents. Where everyone regularly has fun on ChatGPT asking all sorts of questions, AI agents help to structure the completion of a task, whether it involves accounting, data, emails or human resources.

Benkel and co-founder Thomas Friederich had often led workshops and other sessions on digitalisation for the Chamber of Commerce and other players. It was there that they became aware of the real needs of SMEs, which in most cases have neither the human nor the financial resources to access this type of technology, which would enable them to free some of their employees from repetitive tasks with little added value.


Read also


AI "can improve a worker's performance by nearly 40% compared to workers who don't use it", asserted an MIT study from late 2023, whereas when they use it outside their tasks, their productivity falls by 19%. "For each of the 18 realistic consulting tasks within AI's capabilities, consultants using AI were significantly more productive (they completed 12.2% more tasks on average and completed them 25.1% faster) and produced significantly higher quality results (more than 40% higher quality than a control group)", stated a Harvard report, which also used the same data covering 758 consultants. Consultants of all skills benefited significantly from the increase in AI, with those below the average performance threshold increasing by 43% and those above increasing by 17% compared to their own scores.

With , companies could inject their own documentation into "their" AI - since the LLM is installed on their facilities - and then have the agents that will operate in that ecosystem. All this at a cost that is difficult to estimate authoritatively, but ranging from roughly €5,000 to €20,000 per year, depending on the company and the AI agents it needs and wants to put in place, according to Benkel. With just one prerequisite: a graphics card. And just one requirement: if the AI agent in a corporate documentation context can overcome the problems of hallucinations or problems in formulating the correct answer, this implies that someone in the company must keep the documentation up to date. Which is no mean feat.

Read the French-language version of this report /