Beelzebub has raised €3 mn in a seed funding round led by United Ventures. The financing brings total funding to €3.3 mn following an earlier €300,000 pre-seed round.
The Milan-based company develops an AI-native cybersecurity platform designed to detect and respond to attacks carried out with automated tools and AI agents. The new capital will support research hiring, international sales expansion and customer acquisition.
Beelzebub plans to open sales offices in Rome and San Francisco by the end of the year. The company is targeting organisations responsible for essential infrastructure and other high-value systems.
The business started in Milan in 2021 as an open-source research project created by computer engineer and security researcher Mario Candela. It was incorporated as a company in July 2025, while its Milan laboratory remains the centre for malware research and product development.
Beelzebub’s platform follows zero-trust and assumed-breach principles. Instead of relying only on perimeter defences, the software is designed to detect attackers after they begin moving through an organisation’s systems.
When suspicious activity appears, the platform diverts the attacker into an isolated environment built for deception and automated analysis. Security teams then receive information about the activity and the malware involved.
The product consists of three main components. Arcangelo acts as the offensive testing layer, simulating targeted AI-driven attacks against the organisation’s environment.
Beelzebub Managed provides the defensive layer by deploying LLM-based decoys that imitate systems such as servers, databases and APIs. It can also reproduce cloud infrastructure and operational technology devices to attract attackers already present inside a network.
When an attacker interacts with one of these decoys, the platform is designed to isolate the affected machine and begin response procedures. The decoys are updated to reflect newly disclosed CVEs so dormant attackers can be drawn into revealing their presence.
Arcangelo and Beelzebub Managed operate within the same automated attack-and-defence process. The company said this removes handovers between separate security products during testing and response.
The third component, Caronte, acts as an AI malware analyst. After malicious code is captured, it reverse-engineers the malware and produces an incident report, which Beelzebub says takes roughly one hundredth of the time required by a human analyst.
Caronte can run entirely on-premises for organisations that don’t want sensitive information sent to external cloud analysis services. The broader Beelzebub platform is available both as SaaS and as an on-premises deployment.
More than 60 independent researchers contribute threat intelligence to the platform as new attack methods appear. Beelzebub said this information is used to update its defensive systems and malware analysis capabilities.
CEO and founder Mario Candela said automated attacks are becoming too fast and numerous for human security teams to manage alone. He said the company is building systems that respond at machine speed while adapting to new malware techniques.
United Ventures co-founder and managing partner Massimiliano Magrini said AI is changing both offensive and defensive cybersecurity. He pointed to the technical background of Beelzebub’s founding team and its automated approach to attack simulation and defence.
Beelzebub cited the rising financial cost of cybercrime and tighter European regulation as part of the market demand for its technology. The company said the funding will also support development of new products aimed specifically at protecting AI agents.
