Cisco is reportedly in advanced talks to acquire Astrix Security, an Israeli cybersecurity startup, in a deal valued somewhere between $250 million and $350 million.
The potential acquisition, first reported by The Information, seems to be driven by the security risks that come with deploying AI agents at scale.
Founded in 2021 by Alon Jackson and Idan Gour, both veterans of Israel’s elite Unit 8200 intelligence unit, Astrix has built its platform around the concept of ‘non-human identities’ (NHIs).
In plain terms, these are the API keys, service accounts, OAuth tokens, and machine credentials that AI agents and automated processes rely on to operate inside corporate environments.
As AI agent adoption accelerates, the volume of these identities is growing fast, and most organizations still have very limited visibility into what those agents are actually doing once embedded in their systems.
In discussing the news, Alon Jackson, CEO of Astrix Security, said:
“Many organizations want to integrate AI into every aspect of their work, but they need robust security.
“Our platform ensures that AI agents are not only effective but also secure and properly managed.”
Where Contact Centers Come In
Jackson’s statement carries particular weight for the CX space.
AI agents are now routinely deployed across customer service environments to handle inquiries, process refunds, update account records, and route complex cases.
To perform any of those tasks, they need access to CRM platforms, ticketing systems, and knowledge bases – and that means permissions and credentials that can, if left unmanaged, become serious vulnerabilities.
The risks Astrix targets will sound familiar to those considering AI governance in a contact center context.
Below are some of the main issues surrounding contact center AI agents:
- Over-privileged agents that have more access than their function requires
- Shadow AI where employees spin up unauthorized agents using existing API keys
- Credential compromise that can open the door to data exfiltration or unauthorized workflow
Given that these agents are regularly interacting with customer records, payment details, and case histories, a breach at this layer quickly becomes a customer trust problem.
A Market on the Move
Cisco’s appetite for Astrix reflects how seriously large technology companies are beginning to treat this challenge.
Indeed, Astrix’s customer base already includes Fortune 500 companies and names like Workday, NetApp, Priceline, and Figma.
Revenue grew fivefold in the year before the company’s $45 million Series B in December 2024, which was led by Menlo Ventures through its Anthropic-backed Anthology Fund, with participation from Bessemer Venture Partners, Workday Ventures, and CRV. Total funding stands at $85 million.
That Series B came with a notably confident message from Jackson.
“We didn’t raise this round to sell the company within a year or two,” he said at the time.
“We’re building a company that can lead in a highly competitive market.”
Whether the Cisco talks were already a consideration at that point remains unclear, but the direction of travel was not.
If the deal closes, it would push Cisco’s cybersecurity portfolio into a space that has become one of the loudest conversations in enterprise AI deployment.
For contact centers specifically, bringing AI agents into the customer service stack is no longer purely a technology and workflow decision.
The credentials and permissions those agents carry need the same scrutiny applied to any employee with access to sensitive systems.
Cisco’s reported move on Astrix is a signal that the broader market is catching up to that reality.
