Cisco buys Astrix Security to strengthen AI agent discovery and governance

Cisco Systems Inc. today revealed it’s going to buy the Israeli cybersecurity firm Astrix Security Ltd. in an effort to bolster the defenses of artificial intelligence agents. The startup has developed a platform that’s able to identify, manage and secure AI agents running in production environments.

The company did not disclose the terms of the deal, but The Information reported on the talks last month, speculating that it could be worth in the region of $300 million.

Peter Bailey, senior vice president and general manager of Cisco’s Security Business Group, revealed the acquisition in a blog post, where he pointed to Astrix’s focus on “non-human identities,” such as machine-to-machine connections.

“Since its founding five years ago, Astrix Security has focused on securing the identities and credentials that power modern systems — API keys, service accounts, and OAuth tokens — the very credentials that AI Agents are now using (and abusing) to gain access and execute work at scale,” he wrote. “The addition of Astrix Security brings deep capability to discover and secure every AI agent and non-human identity (NHI), including excessive privileges and real-time threats, enabling organizations to adopt AI securely and at scale.”

Astrix’s software allows companies to create and maintain a real-time inventory of all of the AI agents, MCP servers and NHIs running in their environments, together with context around what they’re doing. This enables security teams to understand the risks associated with them.

Cisco said it wants to integrate these capabilities with its Cisco Identity Intelligence platform in order to strengthen the visibility and context of identities within the broader Cisco Security ecosystem. It will also extend Astrix’s functionality to platforms such as its zero-trust offerings, including Cisco Secure Access, Bailey said. “Customers will be able to discover, authenticate, and authorize agentic identities, as well as detect and respond when they use Cisco Secure Access as well as Duo,” he explained. “This visibility and intelligence also feeds into Splunk (or any SIEM), giving security teams a unified view of agent activity with the context needed to investigate and respond at machine speed.”

Astrix’s platform will provide Cisco with some very specific capabilities, including tools for discovering and governing AI agents. It provides a map of an organization’s entire agentic activities, with tools to adjust the policies governing them so they can reduce their attack surface and avoid compliance violations. Astrix also provides tools for AI agent access and lifecycle management, as well as threat detection and response functionality.

Bailey said Cisco urgently needs to beef up its agentic security. He cited the latest edition of the company’s AI Readiness Index, which reveals that only 24% of businesses today currently have the ability to control AI agents with proper guardrails and live monitoring. In addition, just 31% of enterprises believe they are “fully capable” of securing their agentic deployments.

While AI agents and other non-human identities outnumber human users by around 100:1, they mostly fly under the radar, creating a huge blind spot for security teams, said Astrix co-founders Alon Jackson (pictured left) and Idan Gour (right) in their own blog post. “Joining Cisco means Astrix now has the scale, the reach, and the platform to bring agentic and NHI security to organizations worldwide,” they said.

Cisco is making a concerted effort to position itself at the forefront of agentic security. Last month, it acquired the observability startup Galileo Technologies Inc., which focuses on observing and evaluating AI models, in what it said is an effort to make agentic workforces “more trustworthy and reliable.” At the RSAC 2026 Conference in March, Cisco debuted a number of new features that aim to help enterprises secure AI agents, including a new tool called Duo IAM for managing agentic access to business applications. It also launched an open-source tool called DefenseClaw for scanning AI agents for vulnerabilities.

Photo: Astrix Security