Start with the work, rather than the tool
A useful automation brief begins with a specific task. Describe what starts the work, what information arrives, who makes a decision, and what a successful result looks like. ‘Handle new enquiries’ is a starting point; ‘collect the enquiry, check the required details, and send it to the right person’ gives a team something concrete to design.
Walk through a recent example with the person who does the work today. Note the missing information, repeated steps, and exceptions. Those details matter more than a long list of AI features.
Separate rules from judgment
Some steps have clear rules: a required field is missing, a booking slot is unavailable, or a request belongs to a particular department. Write those rules down first. Consider AI for parts that involve interpreting language or preparing a draft, with a clear review step when a person needs to make the final decision.
Decide what the system should do when it is uncertain. A request for clarification or a handoff to a person is a useful outcome. The process should not depend on the assistant always having an answer.
Choose one useful first version
Pick a narrow workflow and agree on how you will evaluate it. You might track how many enquiries reach the right person with complete information, how often staff need to correct the result, and whether the process is easier to use.
Keep access permissions, sensitive information, and recovery steps in the brief. Before expanding the system, review real examples with the people responsible for the work. The next version should follow what you learn from those examples.