Reco Raises $55 Million to Keep Business AI Agents Safe
Reco has raised $55 million to help keep AI agents safe for large companies. The company announced the round on September 29, 2026. TechCrunch
The financing builds on a $30 million round from February. The extension includes AT&T, which also uses Reco and invested through its venture arm, alongside Forestay and Quadrille.
Reco is led by co-founder and CEO Ofer Klein. The company started with software to map and secure business software and AI tools. It now focuses on what it calls a context graph. Think of it as a live map that shows which AI helper can use which app, whose data it can touch, and what permission it used to get in. Each helper is treated as its own identity, with rules for what access it inherits, what it can pass on, and what limits apply.
Reco points to discovery as the hard task. It said its platform found 21,000 agents inside a Fortune 100 customer that the customer did not know about. Large companies already lose track of apps, app-to-app permissions, automated accounts and connections with too much access. Agents add to the problem because business users, developers and other agents can create them outside central IT control.
The February round was led by Zeev Ventures, with Insight Partners and boldstart ventures taking part. That group was disclosed in April. The September extension adds strategic and cross-stage capital without replacing the earlier lead.
The current round has history. In April 2025, Reco described an additional $25 million raise and said it planned to use the funding to grow its AI-native software offering and to close what it called the SaaS Security Gap with Dynamic SaaS Security. Reco Redseed joined as a new investor in that $25 million round at the time. Calcalistech
Since then, Reco was named a CRN 2025 Stellar Startup and doubled headcount across multiple regions. The company has also been named an OpenAI Select Partner, lists a catalog of more than 270 agents and apps, and says its Factory provides more than 260 app integrations.
In my view, two details help explain the news. AT&T as both customer and investor points to use in production rather than pilot testing. The 21,000 unknown agents and the integration counts also count, because protection only works if it finds what IT misses and connects to enough apps.
The broader context here is a crowded field. Many startups are trying to define agent security. Some check agent behavior during operation or filter prompts. Others manage identities and system settings. Reco is betting the control point is the relationship map. If you can track which agent can act on which data, on whose behalf, and through which app permission, you can stop risky actions before they run and review them after.
In my view, that focus makes sense for large buyers. Filters for AI output have a role, but risk will center on permissions. Old access tokens, overly broad access, shared accounts and chained access through connected apps are already known failure modes. Agents make them change faster. A live map is one practical way to keep permissions current as agents are created, share logins and link tools. The open questions are practical. How up to date is the map, how noisy are the alerts, and how well do fixes fit into identity and ticketing tools teams already use.
