Codenotary launches AgentMon for AI agent oversight

Codenotary has launched AgentMon, a monitoring product for AI agents aimed at enterprise users seeking visibility into security, performance and cost.

The launch comes as businesses deploy more agentic AI systems - software agents that act on behalf of users and applications across internal operations. Codenotary cited a BCG forecast that the AI agent market will grow at a compound annual rate of 45% over the next five years.

AgentMon is intended to give AI operations, security and compliance teams a single view of how agents behave, what resources they use and whether they remain within internal rules. It continuously monitors agent activity across different environments.

The product tracks operational health, communication paths between agents and services, token usage, model selection and inference latency. It also monitors file access, secrets handling and data access patterns that may indicate leakage or policy breaches.

Visibility Gap

AgentMon is designed to address a growing problem for companies adopting AI agents at scale: limited visibility into what those systems are doing once deployed. Organisations want to know whether agents are leaking sensitive data, how much they are costing and whether they are performing as expected.

The system correlates token telemetry, behavioural baselines and data lineage to show how agents operate as a wider network rather than as isolated tools. This is intended to help teams manage AI agents more like distributed computing systems while keeping security, compliance and spending controls in view.

Codenotary said its own experience underscored the issue, citing an error it discovered in the setup of agent workflows that had previously gone unnoticed. It presented the finding as an example of the blind spots that can emerge in agent deployments.

Commenting on the launch, Dennis Zimmer outlined the risks companies are trying to manage as AI agents spread through business systems. "Agentic networks are growing explosively, and with that growth come entirely new categories of risk," said Dennis Zimmer, co-founder and chief technology officer at Codenotary.

He added: "Organisations are now asking critical questions: Are agents leaking sensitive data? How much are they costing us? Are they performing as expected? AgentMon brings clarity to these type questions, giving enterprises the visibility and control they need to confidently scale AI."

Target Users

The product is aimed at chief information officers, chief information security officers and compliance leaders seeking to manage AI use across large organisations. That reflects a broader shift in the AI market, where governance and oversight tools are becoming more important as companies move beyond experimentation into operational use.

Codenotary is best known for software supply chain protection. It says it works with hundreds of customers worldwide, including banks, governments and defence organisations, and focuses on security and integrity across the software development lifecycle.

Zimmer said companies need stronger oversight as the use of agents expands. "Agents are incredibly powerful, enabling entirely new business use cases," he said.

"But without proper supervision, they can introduce significant risk. AgentMon acts as an always-on control plane, helping organizations understand what their agents are doing, what it costs, and whether they are staying within guardrails."