The flurry started in February, though nobody was calling it that yet. Warp, a developer terminal startup, announced something called Oz—a platform to manage coding agents. Okta followed in March with an AI agent framework. Then the releases started coming faster.
By the time June closed, more than fifteen vendors had launched platforms aimed at solving essentially the same problem: companies have lost track of what AI tools their employees are using, how much they're spending, and what data might be leaking out in the process. The simultaneity felt less like coincidence and more like confirmation that enterprise software had a new category on its hands.
Shadow AI, Meet Your Managers
The term making the rounds is "shadow AI." It's the enterprise equivalent of what IT departments used to call shadow IT—unauthorized software proliferating across teams. Engineers download Claude or ChatGPT to write code faster. Product teams spin up Cursor or GitHub Copilot without running it past anyone. Marketing adopts Jasper. Finance tries something else entirely.
Within months, a 200-person startup might be carrying dozens of AI subscriptions scattered across teams, no centralized tracking, and a growing list of security questions about where sensitive data is flowing. It's a mess, though perhaps a predictable one.
Torii, which sells SaaS management software, launched its AI Management Platform on May 12, 2026, framing the pitch around visibility: Who's spending what, on which models, across which projects? The platform promised to answer those questions while giving companies a handle on both cost and risk.
Okta took a different tack that March, focusing specifically on AI agents rather than the broader universe of AI tools. The distinction matters. Most organizations don't treat agents—autonomous software that can take actions on its own—as distinct identities within their systems. That creates a security gap, particularly when agents start accessing production environments or customer data. Okta's framework introduced discovery, registration, and centralized management, including what the company called a "kill switch" for agents behaving badly.
Developers Get Their Orchestration Layer
The developer tooling segment moved early, which makes sense given that engineers were among the first to embrace AI coding assistants.
Warp's Oz, unveiled February 10, tackled a specific pain point: coding agents making changes across multiple repositories without much oversight. The platform offers orchestration and governance, running workflows in Docker sandboxes with shared team context—an attempt to bring some order to what can quickly become chaos.
Guild.ai followed April 29 with what it positioned as "the first control plane for AI agents," built for code review, issue triage, and workflow automation. The pitch centered on agent lifecycle management, essentially a layer sitting between raw AI capabilities and actual production deployment.
Coder Agents entered beta around May with a focus on platform teams managing their own infrastructure. The offering lets organizations run AI development workflows on self-hosted systems, appealing to companies unwilling to route proprietary code through third-party clouds. Security-conscious, or paranoid, depending on your perspective.
IT Operations Claims Its Territory
While developers were getting orchestration platforms, IT operations teams were building out their own subcategory: AI-native SaaS management.
BetterCloud announced its "Next-Generation SaaS Management Platform" on June 17, bundling AI governance with an IT agent and centralized visibility tools for enterprise environments. The name is a bit grandiose, but the underlying pitch is straightforward enough.
ConnectWise went broader on June 8, unveiling what it called "the industry's first Predictive Intelligence Platform"—a system spanning professional services automation, remote monitoring, cybersecurity, and agentic AI capabilities. The target audience: managed service providers and internal IT teams juggling multiple systems.
Xurrent launched autonomous AI agents and an open Model Context Protocol server on May 12, targeting IT service and operations teams with an emphasis on integration with existing service management workflows rather than wholesale replacement—a positioning that suggests some wariness about how much disruption customers actually want.
Workplace Productivity Vendors Join In
Atlassian made its move at Team '26 on May 6, opening up its Teamwork Graph via Model Context Protocol to power agentic work across the enterprise and unveiling Rovo Studio for agent design and automation. The company's argument: without a centralized graph and governance layer, agent adoption fragments workflows rather than improving them. It's a reasonable concern, assuming you believe agents will become widespread enough to matter.
Asana positioned its "Agentic Work Management" approach throughout this period—essentially an operating layer meant to prevent the chaos of uncoordinated agent deployment. The core worry: agents without shared context create more silos, not fewer. Whether that's alarmist or prescient remains to be seen.
BasedAI emerged from stealth on May 13 with Hirebase, billing itself as an "instant AI workforce platform" in closed beta. The promise: deploy agents across existing productivity tools without requiring new infrastructure. Ambitious, if nothing else.
Cyndra AI launched its "Secure AI Employee Platform" on June 30, emphasizing human approval loops for agent-generated work. It's a governance model positioned somewhere between full autonomy and manual execution—hedging, essentially, against the possibility that fully autonomous agents aren't ready for prime time.
Testing Platforms Adapt
Quality engineering platforms moved quickly to adapt. mabl announced its "next-generation agentic testing platform" on April 23, framing the challenge as testing at "agentic development speed." The underlying idea: AI-accelerated development breaks traditional QA cycles, so testing tools need to speed up accordingly.
Perforce announced updates to Perforce Intelligence on June 30, including what it called an MCP-agnostic Agentic Gateway, AI-assisted testing, and unified compliance. The positioning targets software delivery and QA teams specifically, particularly those in regulated industries where compliance isn't optional.
Ketryx launched an MCP server beta on March 31 to bring compliance intelligence into AI workflows for safety-critical product development. Think medical devices, automotive systems—domains where governance gaps carry real consequences.
New Entrants Stake Their Claims
Several newer platforms emerged with broader, more ambitious positioning. Osirus AI launched April 21 as a "unified platform for building, deploying, and managing enterprise AI agents." Kore.ai announced its Agent Management Platform on March 17, promising governance across frameworks, clouds, and development environments.
The timing suggests these companies spotted the same opening: established vendors were building AI features into existing products, but nobody owned the cross-platform governance layer. Whether that's a sustainable business or just a gap waiting to be filled by larger players is an open question.
Convergence, Not Fragmentation
This isn't fifteen companies solving fifteen different problems. It's fifteen companies converging on the same insight: AI adoption in enterprises has moved faster than organizations' ability to manage it.
The platforms split along predictable fault lines. Some focus on cost and usage visibility—Torii, BetterCloud. Others emphasize security and identity, like Okta. Developer-focused tools handle orchestration and environment management: Warp, Coder, Guild.ai. Testing platforms adapt to AI-accelerated development cycles—mabl, Perforce. Workplace productivity vendors add governance layers to agent design—Atlassian, Asana.
The CNCF Technology Radar for Q1 2026 noted the trend toward platform approaches for AI workflows, with particular emphasis on security, compliance, and practices for platform teams. Microsoft Build 2026 in May reinforced the agentic AI theme across the development stack.
Perhaps the most revealing detail: most of these platforms didn't wait to establish the category first. They launched into it simultaneously, which suggests the problem had already reached critical mass in their customer conversations.
The Vendor Bet
For engineering leaders and CTOs, the emergence of this category creates a new problem even as it solves an old one. Which layer of the stack do you control—tool-level, agent-level, or workflow-level? Do you need multiple platforms for different use cases, or is consolidation inevitable?
The platforms launching now are betting that IT and engineering organizations will pay for centralized governance rather than trying to build it internally. Whether that bet pays off depends on how painful the shadow AI problem becomes.
Judging by the launch velocity, vendors think it's already painful enough. Whether they're right—or just early—will become clear soon enough.
