Rubrik has expanded its Project Hourglass AI security alliance with a new code-security offering powered by Anthropic’s Claude, extending the program beyond monitoring autonomous AI agents to identifying vulnerabilities in enterprise software before they reach production.
Announced October 9, the expansion introduces Rubrik Code Guardian and adds three solution providers (AHEAD, Trace3, and World Wide Technology) to an alliance that already includes Cognizant, Deloitte, HCLTech, NTT DATA, Wipro, and LTM.
For participating partners, the expansion creates opportunities to help enterprise customers address security risks across both AI-generated code and autonomous agent activity, as organizations move more AI capabilities into production.
“There’s a difference between having security tools and knowing how to put them to work as part of a customer’s AI strategy,” Rubrik Chief Solutions Officer Alok Agrawal told Channel Insider.
Rubrik Code Guardian extends Project Hourglass into code security
Project Hourglass initially centered on Rubrik Agent Cloud, which provides behavioral monitoring, security guardrails, and recovery capabilities for autonomous AI agents.
The addition of Code Guardian extends those protections into software development, allowing partners to help customers identify potentially exploitable vulnerabilities before code reaches production.
Code Guardian uses Claude Mythos 5 within Rubrik’s security framework to evaluate isolated copies of customer code repositories. According to Rubrik, the technology is designed to identify multi-step attack chains, assess exploitability, and prioritize vulnerabilities according to business impact.
“Participants gain access to joint go-to-market resources, technical certification and direct support from our solutions and engineering teams, so they can bring our latest agentic AI solutions to customers faster than they have before without having to navigate integrations on their own,” Agrawal said.
Code Guardian remains in private preview for select design partners and is not yet generally available.
AI agent adoption exposes gaps in enterprise security controls
Rubrik’s expansion comes as organizations increasingly deploy autonomous agents while struggling to establish adequate security controls.
Research published in April by Rubrik Zero Labs found that 86% of IT and security leaders expect AI agents to outpace their organizations’ security guardrails within the next year. The survey of more than 1,600 leaders also found that 88% lack the ability to roll back agent actions without disrupting systems.
Agrawal argued that traditional approaches to cybersecurity do not fully account for the risks posed by agents with legitimate access.
“An agent can operate completely within its guardrails and still take an action that exposes sensitive information or creates new security risks,” he said.
“The old question was ‘Is an intruder trying to break in?’ The new question is ‘Is a trusted agent doing something it shouldn’t?'”
That distinction creates additional responsibilities for MSPs and security-focused solution providers, including monitoring authorized agent activity and preparing customers to recover from unintended outcomes.
What MSPs and solution providers should prioritize
Agrawal also emphasized the importance of helping customers distinguish theoretical vulnerabilities from risks that attackers could realistically exploit.
“A vulnerability that looks serious on paper may pose little practical risk, while seemingly minor vulnerabilities could be exploited together to create a much larger attack,” he said.
For partners, that means moving beyond vulnerability identification toward contextual risk assessments, remediation prioritization, and recovery planning.
Agrawal believes recovery capabilities will become increasingly important as organizations give AI agents greater autonomy.
“Hope is not a strategy in the age of AI. Organizations must assume their agents will overreach,” he said. “Security must shift toward real-time behavioral monitoring and instant recovery.”
READ MORE: We asked Dev Rishi, GM of AI at Rubrik, about the Rubrik Agent Cloud and how it addresses needs for partners and customers alike.