The phrase "AI sprawl" has moved well beyond industry jargon. As enterprises accelerate the deployment of AI agents across their infrastructure, the term now describes a genuine operational headache for security teams. But there is a secondary sprawl emerging alongside it - one made up of vendors rushing to offer solutions to the very problem those agents create.

A scan of company profiles on Crunchbase and PitchBook reveals at least two dozen firms currently selling some variation of AI agent security. Their approaches vary considerably. Some focus on auditing the tools that agents are permitted to use. Others help organizations control which data repositories agents can access. Companies like CrowdStrike are developing detection and response capabilities directly on the devices where agents run. Still others are targeting the problem of unsanctioned AI usage within organizations - finding it, flagging it, and helping teams shut it down.

Despite their differences in technical approach, many of these products make broadly similar promises. Terms like knowledge graphs, continuous monitoring, runtime security, tool access controls, and MCP vetting appear repeatedly across vendor marketing materials. The competitive landscape is crowded, and the messaging is beginning to blur.

Reco Repositions Around a Broader Agent Security Vision

One company navigating this crowded field is Reco, an AI security startup that spent most of its early years selling software designed to map and secure SaaS platforms and AI tools. Over the past year, the company has significantly broadened its scope. It now positions itself around a more comprehensive solution built on what it calls a context graph - a system that links AI agents to the applications, user accounts, people, and permissions they interact with. The goal is to give security teams a clear picture of what any given agent can reach, and the ability to revoke access that falls outside acceptable boundaries.

Reco co-founder and CEO Ofer Klein says the shift was driven by a stark reality: companies are building and releasing AI agents at a pace that outstrips their ability to track them. He cited one Fortune 100 customer where Reco's platform surfaced 21,000 agents the organization had no prior awareness of. In another case involving a large financial services firm, the startup says it identified an agent that had been configured by a former employee. That agent retained the ability to access data within Salesforce and route it to an external domain the company could not monitor.

"The market demand right now for agent security is not only about the agent itself; it's about the entire ecosystem end-to-end," Klein said.

Industry-Wide Concern Over Agent Risk Is Growing

Klein is not alone in framing agent security as an urgent priority. Chris Sestito, co-founder and CEO of security startup HiddenLayer, recently noted that once AI agents reach production environments, the associated costs and risks shift from theoretical concerns to real-world exposure very rapidly. He indicated that more than 50 of his company's customers already have AI agents running in production and interacting with critical systems and sensitive data.

Separately, AI startup Cymphony reported finding approximately 85,000 files at a U.S. public company that had become accessible to AI tools and agents without the organization's full awareness - a finding that illustrates just how quickly data exposure can scale when agent deployments are not carefully governed.

$55 Million Raise Builds on February's Series B

Investor appetite for companies working in this space has been strong, and Reco has moved to take advantage of it. The startup announced on Tuesday that it has raised $55 million in new funding, extending a $30 million Series B it closed in February of this year. The latest round includes participation from AT&T, which joined through its corporate venture arm as an existing customer. Forestay and Quadrille Capital also participated in the extension.

Klein said the company's valuation has more than doubled since the Series B was announced earlier this year, placing it in what he described as the "high hundreds of millions" range, though he declined to provide a precise figure. On the revenue side, Reco's annual recurring revenue currently sits in the double-digit millions of dollars, and Klein expects that figure to triple over the course of this year. The company now counts more than 100 customers on its roster, with financial services firms making up roughly 40 percent of its business.

SaaS Coverage as a Competitive Differentiator

Reco's strategy for standing out in a crowded market appears to rest heavily on the breadth of its existing SaaS integrations. The platform currently connects with more than 280 applications, and Klein says the engineering team can add new integrations within a matter of days. Beyond direct app integrations, the platform also uses browser and network signals to detect agents operating outside of those connected applications - a capability intended to catch shadow deployments that might otherwise go unnoticed. The system includes controls for inspecting prompts and monitoring tool calls made by agents.

The new capital will be directed toward expanding the company's workforce, scaling its sales operations, building out partnerships, and strengthening customer support infrastructure. With this latest raise, Reco's total funding now stands at $140 million.

Whether Reco's combination of SaaS breadth, context-graph architecture, and financial services traction will be enough to carve out a durable position in an increasingly congested market remains to be seen. What is clear is that the problem it is trying to solve - helping organizations understand and control the AI agents operating across their environments - is one that enterprises are treating with growing seriousness, and that security vendors of all sizes are racing to address it.