Mailinblack brings together the AI engines of its cybersecurity solutions under the name Néréus

Under the banner of Néréus, Mailinblack integrates AI engines across its suite of solutions for email protection, password management, and employee training. The Marseille-based company boasts a client base of 26,000 organizations, advocating for an architectural approach in which each component feeds signals to others. This rebranding comes at a time when buyers are increasingly wary of the numerous AI promises from vendors that lack substantiated evidence.

Artificial intelligence has become the entry ticket in the cybersecurity market, where every vendor stakes their claim while buyers demand results. Mailinblack’s reorganization and deeper integration of its offerings align with the evolving AI-powered cybersecurity landscape. This architecture allows for the seamless exchange of information, analyses, data, and alerts, directly impacting the effectiveness of the solution and the provider’s ability to offer a cohesive, maintainable stack without unnecessary complexity.

In this context, Mailinblack, which has been developing cybersecurity solutions for SMEs, healthcare institutions, and government agencies since 2003, has launched Néréus, portrayed as the underlying AI for its four primary products. The company has gradually expanded its portfolio to include email protection with Protect, password management with Sikker, phishing attack simulations through Cyber Coach, and training via Cyber Academy. This shift also includes a move toward indirect sales to target the European market. Prior to this, Mailinblack had already introduced behavioral analysis and machine learning in Protect Advanced, along with a human risk supervision dashboard. Néréus marks a significant step in consolidating these previously independent engines under a single identity.

A Name Built on Two Decades of Software Development

At the end of August, Mailinblack launched a dedicated webpage for Néréus and published a blog post, preceding their official press release. The cited engines, encompassing features like predictive spam email blocking and link analysis upon clicking, were already integrated into its products. “Since 2003, we have independently developed the technological components that form our AI. They are now combined into a single brain that analyzes behaviors, actual attacks, and risk exposure levels for each product, enabling real-time improvements,” stated Thomas Kerjean, Mailinblack’s CEO.

The real innovation lies in the networking of these components, supplemented by a vast data foundation that captures the cyber threats encountered during Mailinblack’s two decades of operation. The company reports that Néréus analyzed 1.9 billion emails in 2025, with 55 million identified as malicious, representing 2.86% of the total volume. These figures position the company among those vendors capable of extracting insights from their own datasets, albeit focused on populations—SMEs and local governments—that larger players may overlook.

Interconnected Signals Enhance Efficacy

Data detected by Protect from email traffic, including malicious senders and domains, is utilized by Cyber Coach to create phishing simulations that mirror the actual attacks the organization experiences, while Cyber Academy uses this information to update its training content. Conversely, risky behaviors identified by Cyber Coach inform Protect’s surveillance, directing it toward potential vulnerabilities. In the Sikker password manager, a module identifies login forms directly within web browsers, ensuring that only aggregated and anonymized risk signals leave the workstation, never individual credentials.

This internal signal circulation is augmented by a network effect; a threat blocked by Protect for one client is neutralized within hours for all other clients, even those that have never encountered the threat. Mailinblack estimates this mechanism leads to approximately one million emails neutralized monthly, claiming its effectiveness is six times greater than random blocking—though this measurement is self-reported and lacks published methodology. Sharing signals among clients is a common practice among major threat protection providers, and Mailinblack’s model is notable for its focus on a homogenous group of medium-sized French organizations.

Proof of Effectiveness Lies with the Vendor

Mailinblack also claims its protection is “unmatched in the market,” a statement that remains unverified as there are no published comparisons with competitor offerings. This assertion highlights the very concern buyers are increasingly attentive to; a July study emphasizes the importance of transparent operational explanations and measurable outcomes as top criteria for evaluating AI vendors. While the Néréus documentation includes detailed mechanisms and unverifiable superlatives, it still falls into the trap of lacking substantial verification.

For the CIO or CISO of a small or medium enterprise or local government, the verifiable components of the offering are best assessed through experience. Simulations that reflect real phishing attacks, updated training content keeping pace with observed campaigns, and a blocking mechanism that improves as the user base grows are tangible benefits users can observe within their own environment. The unverifiable claims, on the other hand, necessitate the same level of scrutiny as the rest of the market, requiring users to test their own email flows, measure false positive rates, and compare outcomes before making a commitment.

Scroll to Top