Security teams have a long list of fixes they know they should make and no time to make them. Axonius has launched a set of virtual workers to take that work off their hands.
Axonius AI Workforce sits inside the company’s Asset Cloud and adds more than 50 types of autonomous workers covering outcomes such as exposure triage, shadow AI and post-quantum readiness. Each worker owns a metric tied to its outcome and acts when an asset drifts from the state it should be in, with results fed into a ranked daily digest.
The launch arrived with the general availability of two July announcements: an AI Companion that takes plain-language questions about an environment and can open tickets or tag devices mid-conversation, and an MCP Server that gives outside agents the same asset data over the Model Context Protocol.
Guardrails before autonomy
Deployment is deliberately staged. Workers start in Off mode, requiring approval for everything, or in Shadow mode, where a team previews what a worker would do before letting it act in Live mode within set limits. Outbound messages and changes to connected systems are held for approval, and each customer runs a private model instance inside its own environment with an audit trail marking every action as human or agent driven.
Chief product officer Moshe Ben Simon makes a data argument rather than a model one: every vendor ships agents now, but an agent acting on one provider’s incomplete view of an asset will be wrong. Axonius underpins the workers with more than 1,400 bidirectional integrations and deduplicated data across 80-plus asset and exposure entity types.
Omdia analyst Theresa Lanowitz said pairing autonomous remediation with a verified view of the environment is what separates agents teams can trust from tools that add noise.
