Runlayer, a Fresh York City-based enterprise AI startup, has recently launched “OpenClaw for Enterprise,” a solution aimed at providing a governance layer for the OpenClaw agent, which has gained popularity since its release in November 2025. This open-source AI tool, known for facilitating autonomous tasks on computers, has attracted both solopreneurs and employees in large enterprises seeking enhanced business automation. However, its rapid adoption has raised significant security concerns, prompting IT and security departments to grapple with the challenges of managing what has been termed “shadow AI.”
OpenClaw operates differently from traditional web-based large language models (LLMs), as its primary agent, previously called “Clawdbot,” often has root-level shell access to user machines. This capability allows it to execute commands with full system privileges, effectively functioning as a digital “master key” without any native sandboxing to protect sensitive data, such as SSH keys and API tokens. The inherent risks associated with this architecture have made it a target for security breaches, especially through techniques like prompt injection, where malicious instructions can be embedded in seemingly innocuous communications.
In a recent interview, Andy Berman, CEO of Runlayer, highlighted the vulnerabilities of OpenClaw, stating that accessing it could be achieved with just 40 messages, leading to complete control within one hour. These risks underscore the urgent need for robust security measures to protect organizations from potential breaches.
The Challenge of Shadow AI
The phenomenon of shadow AI reflects a broader trend where employees adopt advanced tools like OpenClaw for their superior functionality, despite potential organizational risks. Berman noted that the industry has evolved past simple prohibitions against such tools, with employees increasingly connecting these agents to platforms like Slack and Jira, often without IT oversight. This trend has created a complex security landscape where full shell access is granted with little visibility, prompting security experts to caution against the use of such agents.
Runlayer’s Solution: ToolGuard
To address these challenges, Runlayer has developed its ToolGuard technology, designed to introduce real-time blocking with minimal latency. This system analyzes tool execution outputs before they are finalized, effectively catching remote code execution patterns that typically bypass traditional filters. Internal benchmarks indicate that ToolGuard can increase resistance to prompt injection from 8.7% to 95%, significantly enhancing the security posture of organizations deploying OpenClaw.
The suite of tools provided by Runlayer includes OpenClaw Watch, which detects unmanaged configurations and shadow Model Context Protocol (MCP) servers within organizations, and ToolGuard, which actively monitors tool calls to prevent over 90% of credential exfiltration attempts. This proactive approach aims to equip enterprises to govern AI agents effectively, similar to how they have managed cloud services and software-as-a-service (SaaS) applications.
Licensing and Deployment
Runlayer distinguishes its enterprise offering from the open-source community by positioning it as a proprietary commercial layer that meets rigorous compliance standards, including SOC 2 and HIPAA certifications. This focus on security allows organizations to transition from “community-supported” risks to “enterprise-supported” stability, providing essential legal and technical guarantees that large corporations require.
Unlike typical SaaS models that charge per user, Runlayer adopts a platform fee structure based on deployment size and required capabilities, encouraging broader adoption across organizations. This model aims to reduce friction in scaling the security solution, making it more accessible for larger teams.
The Path Forward
As Runlayer continues to enhance its offerings, it is already collaborating with several high-growth companies, including Gusto, Instacart, Homebase, and AngelList, to secure their AI deployments. The response from the market suggests that the future of AI in the workplace will depend not on banning powerful tools but on implementing effective governance structures to manage their use responsibly.
As the capabilities of AI models improve and the costs associated with them decrease, the urgency for robust governance frameworks will only increase. The emphasis will be on enabling organizations to leverage these tools safely, ensuring that security does not arrive at the expense of productivity. For modern Chief Information Security Officers (CISOs), the challenge lies not in restricting access to AI technologies, but in facilitating a secure and governed rollout that maximizes their benefits.
As organizations navigate these changes, the successful deployment of AI agents like OpenClaw hinges on their ability to integrate security measures seamlessly into existing workflows. The focus will be on fostering a culture of responsible AI use, promoting transparency and accountability even as enhancing overall operational efficiency.
