Start with a piece of work that keeps slipping
Think of one useful thing that has been on your list for weeks. It may be a customer note, a new offer, a board paper or a product test.
The problem may not be the size of the job. It may be the effort needed to begin. You need to find the old notes, remember the last choice, shape the first draft and decide what good looks like. A free hour is not enough if the first 45 minutes are spent rebuilding the context.
That is the start cost.
What AI can carry
AI is useful when it lowers that cost without hiding the important choices. It can keep:
- the facts you use often;
- examples of work you like and dislike;
- past choices and the reasons behind them;
- a simple first step for this kind of job;
- a check that brings the work back to you when it is unsure.
This is more useful than asking a blank chat window to “write a plan”. The system starts closer to the way you already think.
A small test
Choose one job that happens at least twice a month. Record the time it takes today, including the time spent finding things and getting ready.
Then build the smallest useful aid for it. Give the AI the source material, one strong example and three checks. Use it on the next real version of the job.
Ask four questions afterwards:
- Did the work start sooner?
- Did the first version give you something worth editing?
- Did you keep the calls that need your judgement?
- Would the saved time be worth more somewhere else?
If the answer is no, change or stop the system. If the answer is yes, keep the context and improve it on the next use.
The larger gain
Time saved is only the first result. The larger gain is what becomes possible because useful work can begin while the idea is still alive.
Pick one delayed job. Lower its start cost before you try to automate the whole process.