Describe the current process.
Choose a task that your team understands well enough to assess. Record what starts it, which information is used, what the output needs to contain and who uses the result.
A useful description might be “prepare a first draft of a meeting summary from approved notes for the meeting lead to check”. That makes the input, output and review role clearer than a broad request to automate administration.
Agree what good looks like.
Decide what would make the result useful before testing. Consider accuracy, completeness, the time required to review it and whether the output fits the next part of the process.
Time saved in drafting can be offset by extra checking. Look at the whole task, including corrections and exceptions, when you assess the result.
Set the boundaries before the test.
Your organisation needs to decide which information can be entered into a tool, who can access it and which decisions remain with people. Review the tool’s suitability against those requirements before using organisational information.
Start with approved examples. Make it clear who checks outputs and what the team should do if information is missing or an answer is unsuitable.
A useful test brief includes:
- The task and expected output.
- The information approved for use.
- The person responsible for review.
- The evidence needed to decide whether to continue.
Plan for the work after launch.
A tool needs an owner. Someone needs to maintain instructions, review changes and help people use it consistently.
Before expanding use, ask the team what changed in the process and where judgement was still required. Use that evidence to decide the next scope and the support it needs.