FireMon launches agentic AI for network security policy

FireMon has launched FireMon Agentic AI for network security policy management, adding AI agents to its existing platform.

The offering is designed to help network security teams investigate access requests, assess policy options and prepare change plans across firewalls, cloud environments and segmentation technologies.

Built into FireMon's current network security management products, it lets customers use pre-built agents, create their own, or connect external agentic workflows through a FireMon MCP Server based on Model Context Protocol.

Network security operations teams often handle access requests that span multiple devices and policy sets. In those cases, engineers must identify the relevant rules, compare technically valid options and prepare proposed changes before an organisation can approve them.

The new AI agents are intended to handle those investigation and planning steps, while leaving approvals and decisions to human practitioners. They operate within assigned instructions, tools and permissions, using policy information already held in the FireMon platform.

"Network security has never been more complex, and the answer isn't more infrastructure. It's putting the policy context FireMon has built over 25 years to work," said Jody Brazil, Founder and Chief Executive Officer, FireMon.

"Customers can deploy our agents, build their own, and connect FireMon to their broader agentic workflows, while practitioners keep the calls that need human judgment," Brazil said.

How it works

One supplied tool, the FireMon Intelligent Agent, can search and analyse policy data, construct and run Security Intelligence Query Language queries, and explain its findings. FireMon said this is intended to let practitioners ask which rules allow access to a service without manually building searches.

Another tool, the Network Designer Agent, is designed to analyse an access requirement and produce firewall policy recommendations. It is intended to help users translate a business request into technical changes for review.

A third tool, the Policy Change Designer Agent, is designed to move an access request through discovery, recommendation and preparation of Policy Plan changes. It can populate required fields such as rule names and pause when a practitioner must choose between valid alternatives.

Proposed work is not automatically deployed to firewalls. Instead, accepted proposals are saved into workflow stages for later review and approval.

External workflows

FireMon also introduced the FireMon MCP Server, which makes selected policy data and supported operations available to external agents. This could allow other workflows to combine FireMon policy evidence with alert data from an incident investigation or prepare policy changes within broader access management processes.

According to FireMon, the MCP-based approach supports both information retrieval and change preparation. External applications direct the interaction and interpret the results, while access is controlled through roles and credentials.

Customers can define agent instructions, assign tools and skills, and restrict permissions to selected device groups. They can also choose a supported large language model, including their own approved model environment.

The AI features are optional and switched off by default. FireMon added that access controls and audit trails are included to support accountability in the workflow.

Operational focus

The launch reflects a broader effort by security software suppliers to apply AI to operational tasks that have traditionally required manual investigation and coordination. In network security policy work, those tasks can be especially complex because a single access request may be satisfied in more than one way, and naming conventions and internal standards vary between organisations.

FireMon's approach centres on preparing recommendations and staged changes rather than making final decisions automatically. Practitioners retain authority over decisions and approvals throughout the workflow.

FireMon said the agentic AI platform and the FireMon MCP Server will be available in November, with the initial offering carrying no additional charge.