XCOR launches to trace outages in minutes. It still pages engineers.

Palo Alto Networks unveiled Cortex XCOR, an AI-driven observability platform, last week, revolutionizing how technical engineers detect and resolve issues. Rather than solely relying on dashboards and manual response, XCOR uses AI agents to autonomously investigate and recommend fixes. The platform's foundation lies in the challenges faced by cloud-native architectures, which previously limited the feasibility of such automation.

Palo Alto Networks acquired Chronosphere in January to develop real-time agentic remediation, allowing XCOR to transform observability from static dashboards and manual remediation to an AI-first experience.

SVP and GM of observability at Palo Alto Networks, Martin Mao, emphasizes that XCOR aims to eliminate the need to wake engineers in the middle of the night to decipher dashboards. The platform operates on a human-in-the-loop model but can be expanded to grant engineering leaders more autonomous permissions over time. While XCOR defaults to a human-in-the-loop model, it can eventually grant engineers a full night's sleep.

Mao foresees a future where SREs evolve from primarily handling dashboard tasks to more strategic architectural work, akin to airline pilots relying on autopilot for smooth operations but still needing experienced pilots to intervene in critical situations. The SRE role will shift towards focusing on higher-level tasks, thanks to the automation of routine root cause analysis and troubleshooting.

Palo Alto Networks' latest offering includes XCOR Operator, an AI assistant designed to help operations teams align AI-coding velocity with contextual relevance. Mao highlights that XCOR Operator understands user intent and serves as a conversational interface for the specialized agents that operate behind the scenes. The platform has demonstrated its ability to solve complex problems, changing the way Mao interacts with the platform and prompting a fundamental rethink in product design.

By granting reasoning models access and capabilities, XCOR Operator enables the discovery of various paths to reach an answer.

XCOR's AI SRE agent autonomously reasons through underlying issues and recommends actions and mitigations in under three minutes on average, with a 75% success rate in complex production environments and a 19% success rate in incidents where analysis was deemed useful. Manual responses, on the other hand, can take up to 20 minutes to identify relevant issues, gather initial context, and locate the appropriate on-call engineer.

Palo Alto Networks aims to reduce response time as users build trust in XCOR's analysis and increasingly automate remediation. The company plans to integrate high-fidelity Real User Monitoring provider Embrace into the XCOR platform, creating a full-stack solution.