Red Hat is transforming Ansible into a control center for AI agents

Red Hat will enhance the Ansible Automation Platform with numerous new features for AI-driven IT automation. A new version is expected in the third quarter, centered around the automation orchestrator. This aims to make the Ansible platform the central execution layer for AI-driven IT operations. The orchestrator analyzes signals from AI systems or agents, recognizes correlations, and suggests actions. The actual implementation is carried out through defined automation workflows in Ansible. Red Hat calls this principle: “AI recommends, humans approve, and the automation platform executes.”

Furthermore, the orchestrator is intended to connect different sources and tools within a common workflow. For example, alerts from IBM Instana, ServiceNow, or Splunk can initiate the same workflow, where a ticket is created, an AI analyzes the incident, and a suitable action is recommended. Subsequently, a person must approve this action before Ansible is allowed to automatically resolve the problem. This means the orchestrator leverages existing automations rather than replacing them. Many companies already have scripts, runbooks, and Ansible playbooks that the new orchestrator integrates into AI-driven processes.

Limit and control rights

“AI agents will only be successful in IT administration if they operate with clearly defined rights within established guardrails,” says Sathish Balakrishnan, Vice President of the Ansible Business Unit at Red Hat. “AI can analyze situations and recommend actions, but execution must always occur through tested playbooks, approvals, role models, and audit trails,” Balakrishnan adds.

Control mechanisms are a focus of the new version. These include role-based access, approval gates, auditing, content signing, and credential management. These control mechanisms are intended to apply regardless of whether an automation is triggered classically task-based, event-driven, or AI-supported.

Ansible 2.7: More transparency and MCP server

In parallel, Red Hat has announced the Ansible Automation Platform 2.7. It brings, among other things, a visual editor for execution environments, a content catalog, and automation dashboards. These dashboards are intended to help better understand the performance and economic benefits of automation.

For AIOps, the Ansible platform has been extended with an MCP server (Model Context Protocol), which builds on existing Ansible controls for user identities, credentials, and role-based access. The MCP server can also be operated in read-only mode to secure risky actions through human-in-the-loop approvals. AIOps Solution Guides have also been added. The first of these offers integrations with IBM Instana, ServiceNow, and Splunk. This is intended for operating environments where observability, ITSM, and security signals already originate from various systems. The orchestrator is intended to transfer these signals into controlled automation workflows rather than replacing existing tools.

For developers and automation teams, there is an extension that allows connecting AI applications and MCP clients via MCP, including Claude or Cursor. Additionally, an intelligent assistant “bring-your-own-knowledge” is supported, which tailors model responses more closely to existing operational knowledge, playbooks, and internal processes.

(mki)