On September 15, 2026, WSO2 announced the general availability of Agent Manager, an open-source control plane designed to bring identity, governance, and scale to the growing army of AI agents enterprises are deploying. The platform, released under the Apache 2.0 licence and deployable as self-hosted infrastructure or managed SaaS, arrives as organizations confront a problem that barely existed a year ago: they are building and launching AI agents faster than they can govern them.
The Agent Sprawl Problem
Gartner projects that the average Fortune 500 enterprise will have more than 150,000 AI agents in use by 2028. Yet as of 2026, only 13% of organizations believe they have adequate governance in place for the agents they already run. WSO2 describes the current state as one where "most organisations have stitched together separate tools for agent traffic, identity and monitoring, leaving gaps that make it difficult to fully see or govern deployed agents."
The problem is particularly acute around agent identity. In most deployments today, AI agents are fitted into identity frameworks designed for human users. There is no policy layer governing what happens when an agent calls a tool, accesses a Model Context Protocol (MCP) server, or interacts with another agent. A compromised or misconfigured agent can escalate privileges, access restricted data, or execute actions its human operators never authorized.
What WSO2 Agent Manager Does
Agent Manager provides a centralized control plane that separates governance infrastructure from agent logic. The key capabilities include:
- Federated management of agents across cloud, on-premise, and hybrid deployments.
- Verifiable agent identity with role-based access control and instant revocation.
- 40+ built-in guardrails enforced at the agent, MCP, and LLM levels.
- Versioned lifecycle management with one-click agent suspension.
- Kubernetes-native sandboxed runtime for isolated agent execution.
The platform is built on open standards including OpenTelemetry for observability, MCP for agent-to-tool communication, and OAuth2 for authentication. It supports popular agent frameworks including LangChain, CrewAI, Amazon Bedrock Strands, and Microsoft Agent Framework.
WSO2 Chief AI Officer Dr. Rania Khalaf framed the design philosophy: "Teams want the freedom to use the best model or framework for the job at hand and control has to respect that heterogeneity. When governance is separated from agent logic, it can scale across frameworks instead of being locked into one ecosystem."
Why Open Source Matters for Agent Governance
The decision to release Agent Manager as open source under Apache 2.0 is strategically significant. Enterprise governance tools are typically proprietary, locking customers into a vendor's ecosystem and pricing model. By making the control plane open source, WSO2 ensures that governance infrastructure can be inspected, audited, and extended by the community.
This matters for three reasons:
1. Trust through transparency. Agent governance requires that organizations can verify exactly what policies are enforced, how identity is managed, and what data the control plane collects. Open-source code provides that transparency. Proprietary governance tools require trust in the vendor's claims.
2. Avoiding vendor lock-in. If your agent governance is tied to a single vendor's platform, switching models, frameworks, or deployment environments means rebuilding your compliance controls. Agent Manager's framework-agnostic design and open-source licence mean teams can switch without losing governance infrastructure.
3. Community-driven guardrail development. The 40+ built-in guardrails represent WSO2's initial contribution, but an open-source community can contribute additional policy templates, identity patterns, and runtime configurations as new agent behaviors and risks emerge.
Industry Recognition
WSO2 was named among notable vendors in Forrester's Agent Control Plane Landscape, Q2 2026 report. The company also won Best Innovation in Open Source AI at the AI Dev Summit's 2026 AI Tech Awards and has co-authored a whitepaper on identity management for agentic AI with the OpenID Foundation. It joined the Agentic AI Foundation, signaling commitment to open standards in agent governance.
What This Means for SMBs
WSO2 Agent Manager targets enterprises managing hundreds or thousands of agents. Most SMBs are not at that scale yet. But the launch carries practical lessons for any business deploying or planning to deploy AI agents.
1. Governance Must Come Before Scale
The 150,000-agent figure from Gartner sounds distant, but the pattern is already familiar. Companies that deploy a handful of agents without identity controls, audit trails, or suspension mechanisms will face the same sprawl problem at a smaller scale. Even a single production agent should have an identity, a permission scope, and a kill switch.
2. Demand Open Standards in Any Agent Platform
Whether you use WSO2, a proprietary tool, or build your own controls, the principles WSO2 has codified should be your requirements checklist: verifiable identity, role-based access, instant revocation, MCP-level policy enforcement, and sandboxed runtime. If a vendor's agent platform does not offer these, you are building on a foundation without guardrails.
3. Open Source Reduces Risk
For SMBs, open-source governance tooling means access to enterprise-grade controls without enterprise pricing. Even if Agent Manager is more than a small business needs today, knowing it exists, what it does, and what capabilities to demand from any agent platform you adopt is valuable.
At Vivid Visuals, we help businesses implement AI agents with governance built in from day one, using the same principles WSO2 has formalized: identity, guardrails, observability, and the ability to suspend any agent instantly.
Read next from the blog : The AI Visibility Gap: Why Your Website Is Not Enough Anymore · Salesforce Koa and the Agentforce ROI Gap: What CRM Buyers Need to Know · Meta's Stilla.ai Acquisition and WhatsApp Business MCP: The Automation Play for SMBs · Google Ads AI Video Dubbing: Why Multilingual Video Just Got Fast
