# Discover and Secure APIs and AI Agents with Harness

This talk explains how enterprises can discover and secure AI agents and the APIs they depend on, arguing that there is no AI without APIs and that API security becomes even more critical as agents move into production. The speaker describes a threat landscape in which API attacks have doubled over the last 12 months, actively exploited API vulnerabilities have risen 43%, MCP-related vulnerabilities have grown 270%, and autonomous agents can execute attacks at machine speed, shrinking the window between discovery and exploitation while organizations still take an average of 55 days to patch. To respond, Harness presents a unified platform for API and AI security that creates a live inventory of apps, APIs, agents, MCP tools, and dependencies; identifies misconfigurations and vulnerabilities across OWASP, API, LLM, and MCP risk categories; supports AI red teaming; and enforces runtime protection against threats such as prompt injection, sensitive data leakage, and abuse. The platform also integrates with Kong and Kubernetes-based instrumentation to expose shadow APIs and AI traffic in near real time.