Most people start with the tools and look for a use afterwards. That is why so many accounts sit unused after a fortnight. The faster route is the opposite: choose one task you already repeat, and rebuild that task with an assistant sitting beside you.
Choose the task on purpose
A good first task is real, boring and yours.
- Real means it has actual inputs — your notes, your numbers, the client's brief. Practice prompts teach you nothing about your own work.
- Boring means nobody minds if the first attempt is rough. The Monday summary, not the board paper.
- Yours means you already know what a good answer looks like, so you can judge the output instead of trusting it.
Settle two things before you paste anything
What your employer allows, and what the tool does with your input. Company data, client material and personal records deserve one deliberate decision rather than a judgement call per prompt. In most assistants there is a setting controlling whether your conversations are used for training; find it now, not later.
The first prompt
I do this task every week: [describe it in plain words]. Below is a real example of the input I start from and the output I produced last time. Ask me up to five questions about how I judge a good result, then propose three ways an assistant could take part of this off me.
Answering its questions is the useful part. You will end up writing down the standards you have been applying silently for years, and those standards become the brief for every later prompt.
Then go one layer deeper
Once one task works, the same pattern repeats: a prompt you trust, then removing yourself from the middle of it with a fixed automation, then building small tools of your own for the gaps nobody else will fill.
Try this today
Name the one weekly task you would most like to shorten, and write down how long it takes you now. That number is how you will know whether any of this worked.