Codenotary has launched AgentMon, a monitoring platform designed to give organisations visibility into what their AI agents are doing in production. The tool tracks agent behaviour patterns, file access, and data handling across systems — addressing a growing blind spot as enterprises move from experimental AI agent projects to real business deployments.
The timing is significant. With industry forecasts predicting that 40% of applications will use task-specific AI agents by the end of 2026, companies are discovering that deploying agents is the easy part. Monitoring them is where the real challenge begins. AgentMon targets CISOs and security teams who need to watch for data leaks, runaway costs, and security policy violations across agent fleets they cannot manually supervise.
The launch sits alongside a wave of AI security tooling announcements. Astrix Security expanded its platform to block unauthorised AI agent deployments across enterprises. Black Duck released Signal, a tool that secures AI-generated code in automated development workflows. Palo Alto Networks shipped Prisma AIRS 3.0 for governing autonomous AI systems. Together, these releases signal that the industry is moving from 'how do we build agents' to 'how do we govern agents at scale.'
For context engineers building agentic systems with Claude Code and similar tools, AgentMon represents an important category emerging in the AI development stack. As agents gain more autonomy — writing code, accessing files, making API calls — the observability layer becomes as critical as the intelligence layer. Teams deploying production agents without monitoring are essentially flying blind.
