The expanded partnership builds on an existing collaboration that already manages API traffic and threat detection for joint enterprise customers. By moving beyond traditional API gateways into the AI infrastructure layer, the new integration offers automated inventory of AI assets, including Model Context Protocol (MCP) servers and LLM-powered services. This allows security teams to track AI resources without manual documentation or reliance on static configuration files.

Beyond discovery, the system provides real-time behavioral analysis designed to handle the non-deterministic nature of AI agents. The platform monitors for specific threats like prompt injection, data exfiltration, and jailbreaking attempts. According to Rahul Sood, GM of Application Security at Harness, this approach embeds security intelligence directly into the connectivity layer where AI traffic flows, addressing the 'shadow AI' blind spots that traditional tools often miss. The integration is available immediately for all joint customers, providing a unified control plane for both legacy API traffic and modern agent-to-agent communication.