ManageEngine Extends Identity Governance to AI Agents for Its 35,000 Customers With SecureAI

AI agents are moving from experiments into daily business processes, but the market has not settled on one way to govern them. Identity vendors start from ownership and access, while security providers emphasize monitoring and policy enforcement. AI platforms build controls around their own ecosystems. The requirement is still the same: organizations need to know which agents are operating, who is accountable for their actions, what they can reach, and how to stop them from acting outside policy.

ManageEngine introduces SecureAI for agent governance

ManageEngine introduced SecureAI to extend identity governance practices to AI agents. Agents need what employees and service accounts already have: a named owner and a record of what they did.

SecureAI is designed to answer three immediate questions.

• Which agents are in the environment?
• Who is responsible for them?
• What are they allowed to do?

It combines agent discovery, ownership assignment, access reviews, risk scoring, and a gateway that evaluates requests to tools and data. The result is a governance layer for organizations that need visibility and control before agent use expands beyond what security and technology teams can track manually.

What SecureAI does

SecureAI discovers agents from Microsoft Copilot Studio, Microsoft Foundry, Zoho’s Zia Agent Studio, and Google’s Gemini Enterprise Agent Platform. It identifies the Model Context Protocol (MCP) servers and AI models connected to the same environment, flags agents without a human owner, and lets administrators assign one or more owners.

Once an agent is identified, administrators can group its entitlements into templates and run access certification campaigns. Reviewers can approve or revoke access with context from a connection graph that shows the path from the agent’s identity and role to a resource. SecureAI also assigns risk scores to AI accounts, mapped to NIST. Its findings map to the European Union Artificial Intelligence Act, ISO/IEC 42001, and the NIST AI Risk Management Framework.

The gateway adds enforcement

SecureAI routes MCP traffic through a gateway that evaluates each request by caller, time, location, and tool. It can allow, redact, or block the request, and its built-in policies follow the OWASP Top 10 and MITRE ATLAS. ManageEngine demonstrated different access levels for marketing and sales agents, along with data loss prevention rules for sensitive information such as Indian national identification numbers.

The gateway is the part of SecureAI that can stop an agent from reaching a restricted system or exposing protected data. The limitation is that the control works only when traffic passes through the gateway. If an agent connects directly to a tool, the gateway cannot enforce the policy.

The product also needs to provide a complete record of what happens after access is approved. Current examples emphasize gateway decisions such as Allow, Redact, and Block. Buyers should ask ManageEngine to demonstrate whether SecureAI records the agent’s subsequent actions, not only the decision made at the gateway.

ManageEngine’s installed base is its main opening

ManageEngine’s clearest opportunity is with its existing customers. These organizations already have a relationship with the company and may prefer to add agent governance through a provider that understands their technology environment and security priorities. ManageEngine says it serves about 35,000 customers, primarily midsize enterprises, and that close to 1,000 customers participate in the advisory board shaping the next phase of SecureAI.

The company is targeting security and technology leaders who need an inventory of agents, named accountability, evidence for auditors, and policy enforcement across Microsoft, Google, and Zoho environments. ManageEngine points to two signals. It says 15,000 to 20,000 employees at its parent company, Zoho, use agents daily, which shows internal use rather than customer demand, while customer health checks have found agents operating without a governance framework.

The best fit is an existing ManageEngine customer who wants to add agent governance through a familiar provider. Organizations using unsupported agent platforms or relying heavily on browser-created agents should confirm coverage before buying.

Providers are approaching agent governance from different categories

An agent governance market already exists and is growing, extending out of AI governance, risk, and compliance (GRC) tooling and AI control centers. Identity vendors, security providers, and AI platform companies are also extending their existing strengths into the problem. Microsoft Entra Agent ID gives agents distinct identities, sponsors, access reviews, and conditional access. SailPoint Agentic Fabric brings agent inventory, human ownership, and access policies into SailPoint’s identity governance platform. AI platform providers also offer controls that are strongest inside their own ecosystems. Against Entra Agent ID and Agentic Fabric, SecureAI’s differentiator is its integration with ManageEngine AD360 and its embedded ManageEngine user base. Integration with identity governance and data governance is common across vendors in this space, and most AI GRC suites already tie risk scoring to MITRE ATLAS and the European Union Artificial Intelligence Act, so neither is unique. The AD360 link was added in a build released October 22, 2026, after launch, and ManageEngine has not publicly confirmed SecureAI integrations with PAM360, Log360, or outside security event monitoring tools. Buyers should compare its gateway enforcement with what the other two offer.

The right choice depends on the environment

The deciding factors include the identity and security tools already in place, the agent platforms that must be discovered, the depth of activity logging required, and whether policy enforcement can cover every route an agent uses to reach tools and data. Buyers should also expect fatigue. In recent client conversations, we have heard complaints about “agentic governance overload,” with vendors from identity, security, GRC, and AI platform backgrounds all positioning themselves as the vendor of record for agent governance. Any vendor in that position competes with products from well outside its own category, not only its direct peers.

An incumbent provider that already runs the organization’s identities, systems, or access processes starts with an advantage, whether that is ManageEngine, Microsoft, or SailPoint. That familiarity can reduce implementation effort and make agent controls part of an existing program rather than another isolated security tool.

Our Take

Organizations should not wait for the agent governance market to mature before building an inventory and accountability model. Knowing which agents exist, who owns them, and what they did is becoming a basic control requirement.

SecureAI is a sensible starting point for existing ManageEngine customers who want to operationalize AI governance through a vendor they already run, with AD360 as the one confirmed link to ManageEngine’s identity tooling. Outside that base, it is one of several entrants in a growing market that clients increasingly describe as crowded, and its security integration beyond AD360 is not yet publicly confirmed. Buyers should still require proof that SecureAI can discover their full agent estate and explain its risk scoring.

Before a pilot, test the two gaps that matter most: Ask ManageEngine to show an agent connecting directly to a tool and to produce the record of what that agent did after access was approved.

One global question remains: Buyers should ask whether and when SecureAI will support regional models such as Mistral and NTT’s Tsuzumi, beyond the Microsoft, Google, and Zoho platforms it covers today.

Want to Know More?

• Securing Agentic AI and the New Security Reality
• Why Cisco, Why Now: How Cisco’s RSAC 2026 Strategy Reframes Agentic AI Risk for CISOs
• AWS Context Turns Enterprise Knowledge Into Reusable Agent Intelligence
• Varonis Atlas AI Security Platform – End-to-End AI Security and Governance Platform
• CyberArk Positions Machine Identity Security as Its 2026 Growth Engine