€28m · Series A · AI · Paris, France
Paris-based Arlequin AI has raised a €28m Series A to develop a new AI architecture based on topological neural networks and expand its technology internationally. The round was co-led by redalpine and OTB Ventures, with Bpifrance's Defence Innovation Fund participating. Existing investors Vsquared Ventures and 10x Founders increased their stakes, while Xavier Niel and ZEBOX joined.
Founded in 2024 by Hugo Micheron and Antoine Jardin, Arlequin builds AI for organisations that need to make consequential decisions from fragmented data while retaining a path back to the underlying evidence. Its current HuDex platform analyses relationships across documents, transactions, video and operational information. The company says the technology is already deployed with governments and large organisations in four European countries.
Tracing decisions back to source data
Arlequin is positioning its system around comprehension rather than content generation. Its product materials describe workflows spanning financial transactions, communications metadata, forensics, due diligence, logistics and cyber intelligence. In each case, the commercial promise is similar: ingest large volumes of heterogeneous data, reveal connections that are difficult to see manually, and let analysts trace a finding back to the source material that supports it.
That traceability matters because the product is aimed at environments where a plausible answer is not enough. A security team, investigator or auditor must be able to show why a relationship was surfaced and which evidence supports the conclusion. Arlequin says HuDex can be deployed on premises and in air-gapped environments, which turns data control and security architecture into part of the product rather than a separate compliance exercise.
The architecture still has to prove its advantage
The Series A will fund a next step: proprietary models based on topological neural networks, designed to learn not only from individual data points but from multi-way relationships across connected datasets. Arlequin says this could expose structures that become harder to detect as datasets grow, while using less compute than architectures built mainly through scale. Those are development goals, not yet demonstrated commercial outcomes.
The investment case therefore joins research and deployment. OTB Ventures argues that Arlequin offers a distinct route for critical decision-making, while the company says it has around 50 employees, including 35 engineers and 15 PhDs, researchers and data scientists. Capital will expand that team, accelerate European and global deployments, and support a planned Silicon Valley AI lab by the end of 2026.
The commercial test is whether Arlequin can turn its architectural thesis into repeatable customer outcomes. Government and enterprise deployments provide demanding real-world data, but they also bring long procurement cycles, integration work and stringent security requirements. If the new models improve the speed or quality of evidence-backed analysis without sacrificing auditability, the company could extend one platform across several high-stakes workflows. If each deployment remains heavily bespoke, scaling the technology will depend as much on implementation capacity as on model performance.
Sources checked: Arlequin AI; OTB Ventures; Tech.eu Funding Explorer.
