HPE Zerto's Bedrock agentic system cuts support cases by 10%

HPE Zerto, a company specializing in cyber resilience and disaster recovery, integrated Amazon Bedrock to create an agentic troubleshooting system that operates entirely on-premises. This architecture choice was driven by strict data residency and latency requirements typical of disaster recovery environments. The system uses a hub-and-spoke multi-agent design, with an orchestrator delegating tasks to specialized agents for Zerto components, all running locally via the Strands Agents SDK.

The results so far indicate strong adoption and tangible benefits: over 20% of customers have embraced the system since its release, and support cases dropped by 10%. This suggests that AI-assisted troubleshooting is not just a novelty but delivers measurable operational improvements. The success hinges on maintaining the delicate balance between autonomous multi-agent investigation and strict data governance, ensuring that agents remain grounded in live data without compromising security.

Going forward, the test will be whether this on-premises model can scale as HPE Zerto expands its agent's capabilities. The company continues to evaluate newer models to improve quality and performance, likely aiming to further reduce the operational burden on administrators while increasing recovery readiness across more complex, multi-site environments.