1. Give a messy problem some structure

Suppose you say, “I’ve got too much to do this weekend.” An AI might offer a generic timetable. Better: share your available time, tasks and priorities — without sharing private details unnecessarily. Ask it to group similar tasks and point out unrealistic assumptions.

Notice the key constraint: two hours. A useful plan should respect it rather than pretending every job fits.

2. Turn projects into small actions

Large goals are easier when you can see the first physical step. “Get fit” is vague. “Put trainers by the front door and walk for 15 minutes after lunch” is actionable. Ask AI to turn a broad goal into a few steps that fit your circumstances.

3. Ask it to challenge the plan

This is the difference between an agreeable chatbot and a useful planning tool. Ask: “What’s missing? Which step depends on something I haven’t done? What could go wrong?”

For an event, it may flag venue confirmation, accessibility or weather. For a work project, it may point out owners, approvals or deadlines that haven’t been agreed. Check any factual requirements yourself.

4. A checklist is often better than an essay

Use checklists for recurring jobs: preparing for a journey, reviewing documents, setting up a home workspace or planning a conversation. Make each item something you can physically complete or verify.

5. Don’t confuse a plan with an action

An AI that prints “I’ve added this to your calendar” may not actually have calendar access. Unless you can see a confirmed tool action, assume it only drafted the content. Copy or enter items into your real task system, or use an approved integration that confirms the change.

Try it yourself

Pick a small project you’ve postponed. Give AI your time limit and the outcome you want. Request no more than five steps, with the first taking less than ten minutes.

Do the first step. Real-world action is the goal, not a beautiful plan you never use.

Can you explain it back?

Answer each question before opening the explanation. You aren’t being graded — this is how you make the knowledge stick.