Arlequin AI announces €28 million funding round

On September 10, Arlequin AI announced a series A funding round amounting to 28 million euros. This funding is intended to support the growth of the French deeptech start-up, which aims to leverage investor interest as a confirmation of the potential of topological neural networks developed by the company.

The company, already deployed in France, the United Kingdom, and Germany, plans to use this funding to accelerate its operations. A key objective is the establishment of a laboratory in San Francisco, focusing on the security of advanced AI systems.

The start-up is led by two distinct profiles. One co-founder comes from political sciences, while the other is a researcher in data science at CNRS, specializing in dimensionality reduction, a mathematical field that is central to Arlequin AI's architecture.

The name of the company reflects the merging of these two worlds. Arlequin, a character from Commedia dell’arte, symbolizes moments where truth and falsehood invert, resembling the informational world the company aims to clarify amidst disorder. The founders believe that the rise of neural networks can address the shortcomings of current mathematical models in understanding human behavior.

The company emphasizes that it does not operate within the realm of generative artificial intelligence, which has been overly simplified to a single family of technologies known as large language models (LLM). The founders assert that this misunderstanding hampers the overall comprehension of the sector.

Hugo Micheron, one of the founders, highlighted that while LLMs are interesting for generation and reasoning, they were never intended to serve as systematic analytical models without bias. He pointed out that in industries such as security, defense, and finance, reliability is paramount, stating that a missed piece of information could have dramatic consequences.

To address the limitations of traditional AI models, Arlequin AI is exploring topological neural networks (TNN). Unlike traditional methods that typically establish two-way relationships, TNN utilize mathematical structures derived from topology to represent complex interactions among multiple elements simultaneously. This innovative approach seeks to enhance the modeling of group dynamics and complex systems, which is crucial for the sectors in which Arlequin operates.