Access and human review for AI Dots

Give the task a human owner

A Dot helps with work, but an extra tool does not resolve unclear responsibility. Choose an employee who understands the instructions, evaluates results and reports changes.

• Who decides whether the result is useful?
• Who answers questions about the work?
• Who decides when an exception arises?
• Who can stop the task when necessary?

Match access to the assignment

We identify the information required and discuss configuration with your administrator. A task involving one folder does not automatically need access to all business information.

Available permissions and settings differ between environments. We assess the setup that can be implemented, document the choices and check them during the practical trial. Your business policies remain the starting point.

Make review points concrete

A general instruction such as “be careful” is difficult to evaluate. Concrete working agreements are clearer:

• Prepare a draft before sending an external message.
• Report missing information instead of filling it with an assumption.
• Identify the source for important case information.
• Escalate an assignment outside the agreed task.

We agree which steps can run independently based on the task and available capabilities.

Test the assignment’s boundaries

We include examples with missing information, conflicting instructions and requests outside the scope. Your team can then assess how the workflow handles exceptions as well as normal results.

After setup, these agreements still need attention. Read our maintenance guide and implementation steps for a practical approach.