Introduction
Abnormal AI, Inc. — rebranded from Abnormal Security in April 2025 — builds a behavioral security platform that connects to Microsoft 365 and Google Workspace through APIs. Co-founders Evan Reiser and Sanjay Jeyakumar started the company in 2018 with a focus on attacks that look legitimate in content but abnormal in behavior: vendor impersonation, payroll diversion, and account takeover without malware attachments.
The company’s public materials now use the Abnormal AI name and abnormal.ai domain. Brel retains the historical slug abnormal-security for URL continuity while displaying the current legal and brand name in the profile title.
What the company does
Abnormal AI models how people, vendors, and systems normally communicate inside an organization, then flags deviations as risk signals. The platform covers email attack detection, identity-related anomalies, insider threat indicators, and controls aimed at unsanctioned or risky use of AI applications in the enterprise environment. Deployment is API-based rather than traditional gateway routing, which reduces mail-flow complexity for IT teams already standardized on cloud productivity suites.
Who it serves
Abnormal AI sells primarily to enterprise security organizations using Microsoft 365 or Google Workspace at scale. Buyers are often teams responsible for email security, identity protection, and security operations center workflows that previously relied on legacy secure email gateways tuned for attachment and URL scanning.
Company background
Founded in 2018 in San Francisco. Evan Reiser serves as chief executive officer; Sanjay Jeyakumar is co-founder. On April 16, 2025, the company announced it would operate as Abnormal AI, Inc., reflecting expanded scope beyond email into behavioral identity and AI governance. The company remained privately held as of this review.
Product and AI capabilities
Abnormal AI uses machine learning over communication and identity metadata — who emails whom, when vendors change payment instructions, whether login patterns shift — rather than relying solely on static threat intelligence feeds. This behavioral layer is the company’s core AI use case. Product expansion into AI tool governance addresses a newer risk surface: employees connecting generative AI assistants to corporate data without security review.
Key developments
2018: Company founded. 2020s: Customer growth among large enterprises using cloud email. April 16, 2025: Official rebrand to Abnormal AI, Inc. and abnormal.ai domain. Post-rebrand: Product marketing emphasizes identity, insider threat, and AI governance modules alongside email protection.
Why it matters
Abnormal AI sits at the intersection of email security and identity security — two budgets that historically sat in different teams. Its rebrand signals that buyers increasingly evaluate messaging security as part of identity behavior rather than as a standalone gateway appliance. For Brel readers, the company is a useful contrast to secure email gateway incumbents and to endpoint-led vendors in the same cybersecurity sector.
Sector context
See Brel’s AI in cybersecurity hub. Related profiles: CrowdStrike (identity extensions), Proofpoint (email security), and founder profile Evan Reiser.
Sources and references
Abnormal AI official about page and April 2025 rebrand announcement.