Network security policy management company FireMon LLC today unveiled artificial intelligence agents for its platform designed to handle much of the legwork behind firewall policy changes.

The release, called FireMon Agentic AI, is aimed at a workload that grows with every firewall vendor and cloud environment an enterprise runs. One business access request can span several devices and policies, often with more than one technically valid way to grant it.

Engineers have traditionally worked through those decisions by hand. FireMon argues the manual coordination ties up expertise while the business pushes for faster decisions.

In the new setup, agents handle the investigation and planning steps, operating inside limits the customer sets on which devices and tools each one can touch. The data behind that work is the same policy record FireMon’s human users already query. Choices that need organizational judgment go back to practitioners.

FireMon supplies three agents of its own, and customers can customize those or build new ones inside the platform. The FireMon Intelligent Agent answers plain-language questions about policy, such as which rules allow access to a given service, generating the underlying queries itself. Firewall rule recommendations for a specific access requirement come from the Network Designer Agent.

The Policy Change Designer Agent goes furthest, carrying an access request from discovery to staged Policy Plan changes. Required fields such as rule names are filled in automatically, and the agent pauses whenever a practitioner has to choose between valid alternatives. Nothing is saved until someone reviews and accepts the proposal. Even then, acceptance only stores the work in FireMon’s workflow, with no automatic deployment to firewalls.

Outside agents get access too, through a new Model Context Protocol server that could let an incident investigation workflow, for example, pull FireMon policy evidence in alongside alert data. Authorized workflows can go further and turn selected recommendations into staged Policy Plan changes. Roles and credentials govern what each external application can reach.

Network security “has never been more complex, and the answer isn’t more infrastructure,” said FireMon founder and Chief Executive Jody Brazil. He said the aim is to put the policy context the company has accumulated to work, with practitioners keeping “the calls that need human judgment.”

AI enablement is optional and off by default. Customers can pick from supported large language models, including one hosted in their own approved environment, and scope each agent’s permissions to particular device groups. Activity is logged to an audit trail.

Both FireMon Agentic AI and the FireMon MCP Server are due in November. FireMon is not charging extra for the initial release.

Brazil started the company in 2004, and the Lenexa, Kan.-based business now reports more than 1,800 enterprise customers in upward of 70 countries.

Image: FireMon