Onyx Security integrates governance controls with Glean’s AI agent platform
Onyx Security says its Glean integration inventories agents, monitors access and applies runtime controls, though availability and pricing were not disclosed.
By Wei-Lin Zhao · AI Correspondent
· 3 min read
Onyx Security’s Glean integration, reported Nov. 13, is intended to add governance and security controls around AI agents built and operated on Glean’s enterprise platform. Neither company disclosed availability, pricing, customer deployments or independently measured results, leaving the announcement as a product claim rather than evidence of reduced risk.
The division of labor is clear in the companies’ description: Glean supplies the environment for building, deploying and running agents, while Onyx supplies what it calls a secure AI control plane. For enterprise teams, the proposed integration is aimed at the operational problem of keeping track of agents that can reach company data, call tools and take actions.
What does the Onyx Security Glean integration do?
According to the companies, Onyx can create an inventory of Glean agents and map the access and capabilities associated with each one. Organizations can then attach an owner, business purpose and risk classification to individual agents.
Onyx also says its controls can apply enterprise governance and compliance policies, then continually assess changes to an agent’s permissions, tool access and data access. During operation, the platform is intended to evaluate prompts, responses and agent actions. The companies say it can block unsafe requests or redirect actions that fall outside a company’s policy. They also describe centralized audit evidence spanning AI environments.
Those functions describe an agent-governance model with five parts: finding the agents in use, assigning an accountable identity or owner, defining permissions, enforcing policy while an agent runs, and retaining records for later review. Onyx’s own definition distinguishes this governance program from compliance: governance determines the internal rules and accountability, while compliance concerns conformance with an outside regime.
The distinction matters as agents move beyond enterprise search and chat interfaces. In an August interview with Techstrong TV, Glean Chief Security Officer Sunil Agrawal said the company had developed from enterprise search into an assistant built on enterprise data and then toward an agentic platform. He said more capable agents may invoke tools and act on enterprise context, creating a broader control problem than a read-only chatbot.
What remains undisclosed
Onyx says its control plane works across SaaS, cloud, endpoint and code environments, but the companies have not provided technical architecture for the Glean connection in the supplied material. They also have not said whether the product is generally available, how its policies are configured, what actions require human approval, or how customers will be charged.
Onyx co-founder and chief AI officer Gil Elbaz said the combination would give customers consistent control of agent activity. Glean Vice President of Partnerships Zubin Irani characterized context and governance as prerequisites for useful, trustworthy enterprise AI. Both are company statements, and the available evidence does not independently test their claims.
The partnership details come from a single news report. The supplied first-party materials from Onyx and Glean provide general context on their AI-security and agent-platform positioning, but do not independently confirm this specific integration in usable detail. For security leaders, it is another example of agent controls being packaged into the wider enterprise security program, alongside access management, monitoring and policy enforcement.
This story draws on original reporting from SiliconANGLE.
