Rein Security raises $25M to secure AI agents at runtime

Why it matters

Rein's round backs a specific security bet: monitor and control agent behavior at execution time. Its reported growth multiples lack underlying revenue and customer figures, and its $35M total does not reconcile with the $8M seed and $25M Series A it has announced.

Matan Bar-Efrat and co-founder Netanel Rubin built Rein Security to monitor what software does in production. On October 8th, Rein said it raised a $25 million Series A to bring that approach to enterprise AI agents, which can reach business systems and take actions without a person directing each step. Glilot Capital and Sienna Venture Capital co-led the round, with Corner Ventures, Atlacle and RNP Capital Advisors participating, according to the PR Newswire announcement.

Bar-Efrat and Rubin did not begin with agent security. In a founder account, Bar-Efrat said their first idea was cyber-insurance software that would use security data to help insurers assess risk. Insurers showed interest, he wrote, but prospective buyers were slow to pay. The founders changed direction toward a security problem with a more immediate operational cost: teams could not reliably tell which vulnerabilities in running applications were actually reachable or dangerous.

The founders brought experience in both security sales and vulnerability research. Both served in Israel's Unit 8200. Bar-Efrat later worked in European sales and business development at Cyberbit and Elbit Systems, while Rubin led vulnerability research at Check Point and founded cybersecurity startup Holcy, according to CTech's January coverage. Rein applies that combination to agents by inspecting their actions in production and intervening when behavior crosses a boundary.

Security at the point of action

Rein says its platform runs as a sidecar alongside AI applications and agents, where the software executes, rather than sending company data through a gateway or proxy. Rein says it can trace code execution and resource access, monitor API calls and database activity, and enforce action-level controls in real time. Its product site also describes posture and supply-chain security features.

A prompt or model response can look harmless while the agent's next action - opening a record, calling an API or passing instructions to another system - creates risk. Rein argues that security teams need execution context to judge those actions and block them. Rein says its sidecar architecture keeps agent data within the customer's environment; the financing announcement describes the technology as patented. Those are Rein's product claims, not independent measures of coverage or prevention accuracy.

Gartner projects spending on securing AI to reach $4.8 billion in 2027, up 68.7% from 2026. The forecast covers application security, usage controls, governance and gateways, among other categories. Rein is pursuing the application-security and action-control parts of that market, where its runtime approach competes with products from security startups and larger vendors.

Rein cites two examples of agent risks, both reported by the company. Its research group, Agent Breakers, presented work at Black Hat USA 2026 describing a compromise of a shopping agent at a top-five U.S. retailer. Rein also says it detected and blocked a prompt injection hidden in a PDF opened by an enterprise onboarding agent. The announcement does not provide independent validation or technical detail sufficient to assess the defenses in either case.

Growth claims and the funding math

Rein says revenue rose eightfold and its customer base grew fivefold since its January 2026 launch. It gives no starting or ending revenue, customer count, or timeframe beyond that launch-to-announcement span, so the multiples show direction without revealing scale. Rein also says it protects thousands of agents that execute millions of actions. Named users include Flex, Swimlane, Lemonade and Dun & Bradstreet; those names establish customer relationships, not the size of Rein's contracts.

The funding total needs similar care. Rein says the Series A brings its funding to $35 million. Its January 2026 launch announcement described an $8 million seed round led by Glilot. Adding the two disclosed rounds yields $33 million, leaving $2 million of the stated total unexplained in the announcements. Rein did not disclose a valuation with the Series A.

The $25 million gives Bar-Efrat and Rubin room to expand product work, agent-security research and hiring, the announcement says. Rein is betting that enterprises will buy controls at the execution layer as agents take on more consequential work. The founders' earlier pivot showed a willingness to change course when buyer interest did not become spending. The new round gives Rein time to prove runtime visibility can become a budget line in a market Gartner expects to consolidate.