Axonius launches AI agents that can remediate security issues

Axonius has launched AI Workforce, a set of virtual workers designed to fix security issues across a company’s connected systems. The useful distinction is that teams can preview actions or require approval before letting workers make changes on their own.

Watch Desk analysis

What happened

SiliconANGLE reports that AI Workforce is generally available to Axonius customers. The workers cover more than 50 security tasks, including exposure triage, shadow AI and post-quantum cryptography readiness. Axonius says they use asset data reconciled across more than 1,400 integrations, track metrics tied to their assigned outcomes and surface results in a daily digest ranked by urgency.

Teams can choose how much autonomy to allow. In Off mode, workers wait for approval; Shadow mode previews what they would do; Live mode lets them act within set guardrails. Axonius says outbound messages and changes to connected systems still require approval. Each customer gets a private large language model instance in a dedicated environment, with an audit trail distinguishing agent actions from human ones.

Two related tools announced in July are also generally available: AI Companion can answer plain-language questions and take actions such as opening a ticket, while the MCP Server lets outside AI tools and agents access asset data.

Why it matters

Security teams can have plenty of alerts and still struggle to follow through. Axonius is pitching agents that move from identifying an issue to carrying out remediation, using a shared view of assets rather than a collection of disconnected snapshots. The preview and approval settings offer a practical way to test that promise without immediately handing every task the keys.

Our read

The compelling part is not that a security product has acquired an AI label. It is the combination of a defined job, a preview mode and an approval gate for sensitive actions. Start by seeing what Shadow mode proposes; autonomous fixes are only helpful when the underlying asset picture is dependable. The launch describes the controls, but customer experience will show how well they work in practice.

What to watch

• Which tasks teams trust to Live mode, and which remain approval-only.
• Whether the asset data is accurate enough to support reliable remediation.
• How customers can monitor AI-credit use and the costs of deploying the workers.

Discussion spark: Would you let an AI agent make security fixes automatically within guardrails, or should every change to connected systems require a person’s approval?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.