Level 1 · Prompt Engineering

Ask for a Specific Output You Can Actually Use

Better AI answers start with naming the thing you need next: a table, checklist, draft, comparison, timeline, or simple visual that is already shaped for action.

Start With The Next Step

Most weak prompts ask for information when the real need is a usable next step. “Tell me about vacation planning” invites a general essay. “Make a 5-day packing checklist for a beach trip with columns for item, quantity, and whether I need to buy it” gives you something you can use immediately.

Before you type, ask: what will I do with the answer? If you need to decide, ask for a comparison. If you need to act, ask for a checklist. If you need to send something, ask for a draft. If you need to understand a process, ask for a timeline or diagram. The format is not decoration. It tells the AI what kind of thinking to do.

Name The Shape

Once you know the next step, name the output shape plainly. A “table” organizes details side by side. A “checklist” turns advice into actions. A “draft” gives you words to edit. A “timeline” puts events in order. A “matrix” is just a table for comparing choices across the same set of factors.

Good prompts include both the shape and the parts inside it. Instead of “compare gym memberships,” ask: “Create a comparison table for three gym options. Columns: monthly cost, distance from home, hours, classes, cancellation rules, best for, biggest drawback.” This prevents the AI from choosing vague categories that sound useful but do not match your decision.

Add The Details That Control Quality

A specific output still needs boundaries. Say how long it should be, who it is for, what tone to use, and what to leave out. For an email, you might ask: “Draft a polite 150-word email to my landlord asking for a repair update. Tone: firm but friendly. Include the date I first reported it. Do not threaten legal action.” These details keep the answer from becoming too long, too formal, or too broad.

The main failure mode is asking for too much at once. “Plan my career, rewrite my resume, and tell me what jobs to apply for” usually produces a blurry answer. Split the work into stages: first ask for a short plan, then a resume checklist, then a draft summary. This research-to-outline-to-draft pattern works because each answer becomes the input for the next one.

Ask For Visual Structure When Text Is Not Enough

Some topics are easier to use when they are arranged visually, even inside a normal chat. You can ask for a customer journey map, a weekly meal plan grid, a decision tree, a pros-and-cons spectrum, or a learning roadmap. The key is to specify the stages and dimensions. For example: “Make a simple journey map for buying a used car. Stages: research, test drive, negotiation, inspection, purchase. Include emotions, questions, risks, and next actions.”

Treat the first answer as a draft, especially for charts, diagrams, and dense tables. Ask the AI to check whether the structure matches your goal: “What is missing from this table if I need to make a decision by Friday?” or “Simplify this into three sections I can put in a note.” The goal is not a prettier answer. It is an answer already shaped for the next action you actually need to take.

Key takeaways

  • Ask for the output you need next, not just information about the topic.
  • Use clear formats such as tables, checklists, drafts, comparisons, timelines, maps, and decision trees.
  • Name the exact columns, stages, sections, or categories you want included.
  • Add boundaries: length, audience, tone, deadline, and what to exclude.
  • Break large requests into stages: plan first, then structure, then draft or final output.
  • Revise the format by asking what is missing, unclear, too long, or not useful for your next step.