Run a three-day time log
For three working days, note every task longer than 10 minutes: what it was, how long it took and whether it repeats. Use a simple table or a notes app. Do not change how you work during the log; you want an honest picture.
Spot the best candidates
Circle tasks that match at least two of these criteria. They are your first AI targets.
- Frequent: daily or weekly
- Text-heavy: writing, reading, summarising, replying
- Rule-based: you follow the same steps each time
- Low risk: a mistake is easy to catch and fix
Estimate the gain
For each candidate, estimate time per week now and after AI assistance. A task that takes 3 hours a week and could take 1 hour saves about 100 hours a year. Start with the biggest realistic gain, not the most exciting tool.
Pick three to start
Choose three tasks and ignore the rest for a month. Spreading your effort across ten small experiments rarely produces a lasting habit.
Example
A consultant's log
Emails: 1h30/day. Meeting notes: 3h/week. Proposals: 4h/week. Social posts: 2h/week. Chosen first: emails, meeting notes, proposals.
Prompt to try
Replace the [brackets] with your own details and review the result before using it.
Here is my time log for three days: [paste]. Group the tasks, estimate weekly hours for each, and rank them by how much AI could realistically help (frequency, text-heavy, rule-based, low risk). Suggest my top three and why.
Exercise
Log your time for three days and circle three tasks.
