Mechanism
Six steps after approval
Expert view
The gap between decision and reality
In a classic process an approved request lands in the administrator's queue. From that moment the organisation has two "truths": the one in the ticketing system, according to which the access has been granted, and the one in the application, where the access may already be there, may not be there yet, or may have been granted in a different scope because the administrator understood the description differently.
For audit this distinction is crucial. Evidence of a control is not the decision but the consistency of the actual state with the decision. That is why auto-implementation in sara.next always has two steps: execution and verification — checking in the target system whether the expected state has been reached.
Not everything is worth automating
From experience: automation pays off where there are many changes and the application interface is official and stable. Trying to automate a system with no interface for managing permissions — through undocumented writes to the database — ends in a brittle solution that nobody wants to maintain.
That is why the same process also works in a manual model. The administrator receives a precise, approved work order, confirms its execution, and a separate person can verify it. The history is identical; the only difference is who made the change.
Operations
What can be executed automatically
The scope depends on the interface that the target application provides.
| Operation | Example |
|---|---|
| Creating access | Adding a user to a group, role, project, team, workspace or resource. |
| Modifying access | Changing the scope, level or parameters of existing access. |
| Change of user | Transferring access to another person on a role change, cover arrangement or reorganisation. |
| Removing access | Revoking a membership, role or assignment. |
| Execution verification | Checking whether the state in the application matches the decision made in sara.next. |
Results
What changes once it is switched on
Less manual work
The integration takes over the repetitive granting and revoking of access.
Fewer errors
Decisions do not need to be retyped into multiple systems.
Lower handling cost
Requests are fulfilled without involving the application support team.
Proof of execution
The result of the operation is linked to the request and the decision.
Control
Automation without losing control
Automatic execution is always preceded by a request and approval and followed by verification.
Integration modelsSafeguards
- Each application has its own access token, which can be renewed or deleted
- Technical integration accounts have only the scope needed for the operations
- A failed operation goes to technical support or manual verification
- Applications without an interface remain in the manual model — with the same process and history
For technical teams
How to configure it
Once approval is complete, the request goes into a queue that the scheduler processes every minute; for each request a Groovy script is run on an SQL or REST connection. "Every minute" is how often the queue is checked, not a guaranteed completion time.
See sara.next on your own scenarios
During the presentation we will walk through an access request, approval, automatic execution of the change and a review — using examples from your organisation.