Quick Facts
- Cohesity Agent Resilience uses point-in-time snapshots to restore AI agent memory, configuration, and the data systems agents manage.
- The product integrates with Amazon Bedrock at launch, with ServiceNow and Datadog partnerships also announced; Microsoft and Google platforms are on the roadmap.
- General availability is targeted for the end of 2026; select customers can access it now through their Cohesity account team.
Cohesity introduced Agent Resilience on September 16 at its Catalyst 2026 virtual event. The new capability lets enterprises discover, protect, and recover the infrastructure behind AI agents when those agents corrupt data, act on bad instructions, or get compromised.
The announcement comes as AI agent deployments accelerate across enterprise software. Gartner predicts up to 40% of enterprise applications will include integrated task-specific agents by 2026, up from less than 5% previously. Yet 56% of organizations said they were not prepared to detect or contain unintended actions by AI agents.
What the Product Does
Agent Resilience has two core functions. First, it protects agent state through point-in-time recovery, restoring agents to a known-good condition after memory corruption or malicious activity. Second, it protects what agents manage, including databases, file systems, and connected services.
An agent topology feature gives teams a unified view of each agent’s memory stores, connected applications, databases, and supporting infrastructure. That view helps teams assess protection coverage and identify what needs to be restored during an incident.
Cohesity CEO Sanjay Poonen said the product applies the same immutable, point-in-time recovery enterprises already rely on for critical data. Chief Product Officer Vasu Murthy put it plainly: “Detection can tell you an AI agent went off course. It cannot undo the changes.”
Partner Integrations
At launch, Agent Resilience supports Amazon Bedrock AgentCore and Amazon Bedrock Agents. Cohesity also announced strategic integrations with ServiceNow and Datadog, both first disclosed in March 2026.
The Datadog integration combines continuous observability with automated recovery. When Datadog detects an anomalous agent action, such as an unexpected drop in object count in cloud storage, it can trigger an API-driven Cohesity recovery workflow automatically. In a demonstration scenario, deleted records were restored from an immutable snapshot within minutes without overwriting unaffected data.
The ServiceNow integration connects ServiceNow’s AI Agent Control Tower with the Cohesity Data Cloud. Organizations building agents natively within the ServiceNow AI Platform will be able to restore those agents and their associated data to a verified point-in-time state. Those integrated capabilities are expected later this year.
Broader Security Push
Cohesity framed Agent Resilience as part of a wider vision it calls Autonomous Cyber Resilience, which uses agentic workflows to automate a five-step framework: protecting data, identity, applications, and agents; ensuring recoverability; remediating threats; and practicing recovery. The goal is to reduce manual work from incident response.
The Catalyst event featured keynotes from Cohesity alongside leaders from Anthropic, Google Cloud, AWS, NVIDIA, and McDonald’s. Cohesity counts 70% of the Fortune Global 500 among its customers across more than 140 countries and counts NVIDIA, Amazon, Google, IBM, Cisco, and HPE among its backers.
Enterprises that have already deployed agents in production workflows face a specific gap: monitoring tools can flag bad behavior, but they cannot restore what an agent has already changed or deleted. That gap is what Cohesity is targeting with this release.
Read more: Cohesity’s new Agent Resilience lets companies roll back AI agents that go wrong