Agentic AI management has come to the fore as one of the primary capabilities that every enterprise IT vendor now claims to offer competencies in. Among the more vocal in this space now is WSO2 with its new WSO2 Agent Manager, an open control plane for AI agents, that provides software engineering teams with a way to identify, govern, secure and scale agents across environments. But in this crowded market, is there anything new on offer here?
Although it sounds like some form of web standard or working group consortium, WSO2 is in fact a company known for its composable 100% open source stack and agent platform for orchestrating autonomous AI agents and delivering API management, integration and identity services.
The organisation says that WSO2 Agent Manager “addresses a critical gap” in that it is designed to bring visibility, control and accountability to autonomous agents as they work in live production environments.
Non-linear agentic productivity gains
According to Rania Khalaf, chief AI officer at WS02, in the rush to adopt agentic services driven by the promise of non-linear productivity gains that agentic systems can unlock, operational maturity is lagging behind.
“Many teams are forced to choose between moving fast with limited visibility and control, introducing significant unmanaged risk, or slowing progress to build operational frameworks for each runtime and environment. According to Gartner, more than 40% of agentic AI projects are expected to be canceled by 2027 due to rising costs, unclear value, and insufficient risk controls.
WSO2 Agent Manager is positioned as an enterprise control layer, making agents “first-class governed participants” in enterprise systems.
Autonomy & probabilistic behaviour
“AI agents introduce a fundamentally new challenge. Their autonomy and probabilistic behaviour make them powerful but also difficult to control,” said Khalaf. “With WSO2 Agent Manager, we’re bringing agents into the enterprise fabric where they are no longer invisible processes, but identified, governed, accountable entities that can be securely operated at scale.”
As enterprises deploy increasing numbers of agents, many are encountering “agent sprawl” with limited coordination, inconsistent controls and growing compliance risks, often compounded by fragmented tooling across frameworks, runtimes, and hyperscalers.
Khalaf promises us that WSO2 Agent Manager addresses agent sprawl by establishing a centralised system of record for all agents, enabling software teams to choose the framework that best fit their use cases while maintaining a consistent approach to agent operations.
This includes unified governance, performance insights, and policy enforcement for agents running both within Agent Manager and across external environments.
Key functions
Central capabilities include federated agent management – to manage agents across frameworks and environments, i.e. cloud, on-premises, and hybrid, from a single control plane. We can also note agent identity and access delegation – to establish strong identity for agents and securely delegate access, ensuring every action is authenticated, authorised and auditable.
Centralised governance and guardrails define and enforce policies across agents, LLMs, and tools to control behaviour, reduce risk, and ensure compliance. A secure, scalable runtime allows developers to deploy agents in a Kubernetes-native, zero-trust environment with isolation and lifecycle controls, including real-time intervention.
Open, framework-agnostic foundation
This technology is built on open standards, including OpenTelemetry, OpenAPI and MCP, enabling compatibility with leading frameworks such as LangGraph, CrewAI, and Ballerina without vendor lock-in.
WSO2 Agent Manager, which will be generally available in June 2026, is released under the Apache 2.0 license and is designed to avoid vendor lock-in. By embracing open standards and an open source license, the platform strengthens digital resilience, giving organisations the freedom to innovate while retaining enterprise-grade control over their agents and full ownership of their technology stack.
What to think about WS02
The casual reader might well see this story and think – oh, gosh, another vendor thinks AI agents need “governing” as the marketplace for agentic governance almost appears to be expanding faster than that for agentic creation.
It may be the case that Agent Manager has arrived “fashionably late” (which could be a negative, or might mean that this is a more considered and well-architected total technology proposition) into a space already known for competing technologies from ServiceNow, HashiCorp, Dynatrace and a whole heap of startups all promising us a control plane for agents. What is pleasing is the Apache 2.0 license and the use of open standards; let’s hope that openness translates to clarity for all users.