Nobody has written down which changes the agent may run without a person.
Network and IT operations
Let the SRE agent make more changes on its own
Routine scale-outs run inside the change window. A risky rollback during business hours goes to a change manager, and a change during a freeze without a ticket never runs.
Connecting the agent? See the developer guide for Telecom & IT Operations
Why agents stay in suggestion mode
Most operations teams let an agent propose changes but not make them.
A change that needs approval waits until someone happens to notice it.
After an incident, the review has to piece together what the agent proposed and who approved it.
A billing rollback at two in the afternoon
The simulator’s second change, with the people it reaches.
Head of operations
Wants the agent handling routine changes so engineers stay on project work.
Change manager
Gets the rollback in the Inbox with the service, runbook, and 8,500 affected customers, and approves or declines it.
Governance engineer
Adds a freeze rule and watches it in Shadow before it applies.
Incident reviewer
Reads the agent’s proposal, the rule, and the approval in one record.
Built-in handoff
Big daytime changes reach the change manager
Change manager
A change affecting 5,000 or more customers during business hours arrives with the service, runbook, and customer count. If it sits, reminders go out and it escalates.
See the six role portals and how work moves between themWhat operations leaders get
Limits you can widen
Start the agent on small changes. As the record shows it staying inside the rules, raise its limits.
Your change types
Rules follow the standard, normal, and emergency change types your ITIL practice already uses.
Reviews that start from facts
The proposal, rule, approval, and outcome sit together for the post-incident review.
Changes worth handing to an agent
Begin with reversible, well-understood changes, then widen the limits as the record builds.
Scale-outs and restarts
Routine capacity changes run inside the window. Freeze rules stop anything without an approval ticket.
Rollbacks and failovers
Changes with customer impact go to a change manager with the numbers attached.
Incident routing
The agent routes an incident to a responder. The routing rule and alert stay together for the review.
Give the agent its first change window
Start with one service and the change types your team already treats as standard.