You open AI because your brain has 47 tabs open. Twenty minutes later, the AI has given you a twelve-part framework, six business models, four morning routines, a content calendar, a meal plan, and a suspicious amount of enthusiasm. Your original problem is still sitting in the corner.
AI is useful when it reduces the thinking load around a decision. It becomes noise when you use it to generate endless possibilities instead of helping you move.
GIVE IT CONTEXT, CONSTRAINTS, AND A JOB.
A vague prompt produces a broad answer. Broad answers are exactly what an overwhelmed person does not need. Before asking, decide what role the tool is playing.
- Context: What is happening?
- Constraints: What time, money, energy, access, or obligations are real?
- Decision: What are you trying to choose or understand?
- Output: What would be useful—a ranked list, three questions, one next move, a draft, or a comparison?
FIVE PROMPTS THAT PRODUCE SOMETHING USEFUL.
Prioritize the mess
“Here is my unsorted list. Ask me up to three questions, then rank the items using deadline, consequence, effort, and what each item unlocks. Recommend one move I can complete today.”
Shrink the project
“This task feels too large: [task]. Break it into visible actions. Identify one useful step that can be completed in 20 minutes without pretending the whole project will be finished.”
Challenge the plan
“Review this plan for unrealistic assumptions about time, energy, money, access, and other people. Tell me what is most likely to break and suggest a simpler version.”
Compare options
“Compare these options using the criteria I provide. Show the trade-offs. Do not choose for me; tell me what question should decide it.”
Draft the awkward thing
“Help me write a clear, respectful message that says [decision]. Keep it under [length]. Do not overexplain, apologize excessively, or invent details.”
ASK IT TO REDUCE, NOT EXPAND.
When you are stuck, do not ask for 50 ideas. Ask for three viable options and the difference between them. Do not request the ultimate guide when you need the next action. Do not ask it to plan your entire year because tonight feels messy.
Useful follow-ups include: “Make this smaller.” “What assumption are you making?” “Which part can wait?” “Turn this into a checklist.” “Give me one recommendation and explain the trade-off.”
KEEP YOUR HAND ON THE WHEEL.
Generative AI can produce inaccurate, incomplete, biased, or outdated information. NIST’s AI Risk Management Framework emphasizes managing trustworthiness and risk when AI systems are used. For everyday use, that means verifying facts—especially anything involving health, money, law, safety, employment, or other high-stakes decisions.
- Do not treat an AI response as a diagnosis, legal opinion, financial instruction, or verified fact.
- Check important claims against reliable, current sources.
- Do not paste confidential work information, passwords, account numbers, private health records, or another person’s sensitive information.
- Review the privacy and data settings of the specific tool you are using.
- Make the final decision yourself—or with an appropriate qualified professional when the stakes require one.
A thinking partner can hold the flashlight. It should not quietly become the driver.
END THE CONVERSATION WITH A MOVE.
Before you open another prompt, write down what changed: What did you decide? What is the next action? When will it happen? What still needs verification?
If the conversation gave you more information but no clearer move, ask AI to summarize the decision in one sentence and identify the smallest useful action. Then close the tab and do the thing.