Manage RPA solutions
Operate the estate
Day-two work centres on machines, machine groups, connections and runs. A machine must show as ready and be assigned to a machine group with sufficient RPA capacity before an unattended run can start. Watch cloud flow run history and desktop flow run details, including console logs, screenshots and recorded video, and build analytics or alerts so failures are noticed before users report them.
Access and governance
Sharing matters: a co-owner can edit, a run-only user can trigger but not modify. Rotate credentials stored in Key Vault on schedule, and review connections so that changing a shared connection does not silently break every flow using it. Data loss prevention policies apply at environment level and cover desktop flows, so a policy change can block a previously working flow. Audit activity through Purview and, for larger estates, use the CoE Starter Kit for inventory and compliance reporting.
Change management
Promote changes through solutions, not by editing production directly. Keep a version history of the desktop flow, and record which solution version is deployed where. When a UI changes, expect every flow touching that screen to need a selector review.
Worked example
An unattended flow that worked last week now fails immediately. Checklist order: (1) is the machine online and in the group; (2) does the group have free capacity; (3) are the connection and credentials valid; (4) are input values present; (5) has the target application UI changed? Working in that order avoids re-authoring a flow that was never broken.
Rule of thumb: never troubleshoot automation logic until connectivity, capacity, credentials and inputs are confirmed healthy.