Arlequin AI
Arlequin AI provides an AI-native platform for critical decision-making in high-stakes environments, using unsupervised systems to analyze large, fragmented datasets without hallucinations or black-box reasoning. The platform maps data from full topology down to individual passages, enabling users to trace every finding back to its source evidence. Arlequin is developing topological neural networks to capture complex, multi-way relationships at scale while requiring significantly less compute than traditional large language models.
- Artificial Intelligence
- Data & Analytics
- Cybersecurity
- Defense Tech
- Government Technology
- Software Only
Funding
Raised to date
€32.4MRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Est. $35.3Macross 2 rounds
+7€28M · Series A
Announced Sep 2026 · from article date
Disclosed cumulative: $35.3M
Founders
Product
Problem
Organizations facing high-stakes decisions—such as counterterrorism investigations, fraud detection, and cybersecurity—must analyze massive, fragmented datasets where errors or unverifiable outputs are unacceptable. Existing AI systems, particularly large language models, generate text rather than understand structure, producing hallucinations and opaque reasoning that cannot be trusted for consequential actions.
Solution
Arlequin AI provides an AI-native platform that performs unbiased, hallucination-free analysis of large-scale, heterogeneous data by comprehending patterns rather than generating text. The platform maps data from full topology down to individual passages, allowing users to explore queries, open new lines of questioning, and trace every finding to the underlying evidence. Arlequin is building proprietary models based on topological neural networks (TNN) that learn from the structure and relationships within data, capturing complex multi-way interactions at scale while requiring significantly less compute than conventional architectures. The system supports document ingestion, transaction tracing, entity mapping, and cross-system event correlation, with results that remain grounded in verifiable source data.
Target Audience
Primary customers are governments and large organizations across Western and Eastern Europe working in security and defense, criminal and fraud investigations, money laundering detection, information integrity, cybersecurity, and AI system safety.
Features
- Unsupervised systems for large-scale data analysis with no black-box reasoning or context-window limitations
- Topological neural network architecture designed to learn from data connectivity and multi-way relationships, not just individual data points
- Full-topology-to-single-passage mapping that traces payments, transfers, contacts, devices, and anomalous routes back to source evidence
- Real-time document scanning and ingestion with throughput exceeding 800 documents per second
- Explore, search, and chat interface for running forensics and surfacing high-risk networks, connected entities, and anomalous activity
- Architecture designed for significantly lower compute requirements, reducing dependence on energy-intensive infrastructure