Move Beyond the Search Box
Most people use AI like Google, typing in a short question and hoping for the best. When you do this, the AI tries to be everything to everyone, which often leads to bland or generic answers. To get better results, you must shift your mindset and view the AI as a new team member with a specific job description.
Think about how you would ask a person for help. You would not just say "Tell me about taxes" to a random stranger. You would find a tax accountant and explain your specific situation. By telling the AI exactly who it is supposed to be, such as a detail-oriented travel agent or a concise executive assistant, you narrow its focus and improve the quality of its logic.
Define the Personality and Background
A job title is only the first step. To make the AI truly effective, you need to describe its personality and its approach to work. If you need someone to help you learn a new skill, tell it to act as a Socratic tutor who asks you challenging questions rather than giving you the answers right away. If you need help with a draft, tell it to be a blunt editor who focuses only on finding logical flaws.
Defining these traits prevents the AI from being too polite or vague. You can even give it a belief system for the task. For example, you might tell it that you believe every email should be under three sentences or that complex ideas should always be explained using analogies from gardening. This level of detail ensures the AI stays on track and delivers work that fits your specific needs.
Provide a Clear Step-by-Step Method
Once you have established the identity, you must give the AI a set of instructions for how to perform its tasks. In professional settings, these are called Standard Operating Procedures. Instead of a vague request like "Help me plan my week," give your scheduling assistant a specific checklist. Tell it to look at your priorities, identify potential conflicts, and then format the final plan into a simple table.
You can create these instructions by thinking about the steps you usually take to finish a job. If you are researching a topic, tell your research librarian to first find the main facts, then find opposing viewpoints, and finally summarize everything into a one-page report. This structure helps the AI move from just talking about a topic to actually producing a finished product that you can use immediately.
Key takeaways
- Start every prompt by giving the AI a specific professional identity, such as a meticulous proofreader or a supportive coach.
- Describe the personality and core beliefs of the AI to ensure the tone and depth of the response match your expectations.
- Create a simple checklist of steps for the AI to follow so that it functions like an employee following a manual.
- Use test me prompts when the AI is in a tutor role to move from passive reading to active memory building.
- Provide specific context about your goals so the AI does not have to guess what you find important.