Reco Raises $55M to Monitor AI Agents, Backed by AT&T

US-based cybersecurity startup Reco has raised $55 million in new funding to build technology that monitors AI agents operating inside enterprise systems. The round includes a strategic investment from AT&T, signaling interest from large enterprise and telecom players in tools that can track, audit and secure autonomous AI activity.

Reco's focus is on visibility into AI agents as they take on more operational tasks across business software — flagging what these agents are doing, where they have access, and whether their behavior introduces risk. The involvement of AT&T as a strategic backer suggests the capability is being evaluated not just as a security product, but as infrastructure for companies deploying AI agents at scale.

As organizations hand more decisions and workflows to AI agents, the question of oversight becomes a core operational concern rather than a technical afterthought. Monitoring tools like the one Reco is building sit at the intersection of security, compliance and trust: they determine whether leaders can confidently scale agentic AI into customer-facing and back-office processes without losing control of what those systems are actually doing.

For digital transformation leaders, this funding signals that the market is maturing past "can we deploy AI agents" toward "can we govern them responsibly." That shift matters for experience design too — an AI agent acting on a customer's behalf, or inside a support workflow, is only as trustworthy as the monitoring layer behind it.

— $55 million raised by Reco in the funding round, with AT&T participating as a strategic investor.

Agent monitoring is the unglamorous infrastructure that agentic AI's credibility will actually depend on. It's easy to get excited about what AI agents can do; it's harder, and more important, to know what they're doing right now, in production, at your company.

Most organizations rushing to deploy AI agents are optimizing for capability and skipping oversight, which is exactly backwards from how trust gets built with customers and employees alike. The behavioral reality is that people forgive AI mistakes far less than human ones, because an autonomous system failing silently feels like a broken promise, not an honest error. Any operator deploying agents into service or operational workflows should treat monitoring and auditability as a launch requirement, not a post-incident fix — visibility is the actual product here, not a compliance checkbox.