Prompting for Different Output Formats

About this module

The module starts with the workplace problem, not the tool. The section uses requesting tables, checklists, emails, briefs, JSON, outlines, summaries, and slide notes to make prompting for specific output formats easier to apply. It also names the common trap: forgetting to define structure, length, fields, or formatting rules. Learners leave with a clear next step, a review habit, and enough context to get AI output that is easier to paste into the next workflow without treating the AI output as finished work.

Key takeaways

  • Explain prompting for specific output formats in plain business language
  • Practice requesting tables, checklists, emails, briefs, JSON, outlines, summaries, and slide notes with the right amount of context
  • Catch forgetting to define structure, length, fields, or formatting rules before the output moves forward
  • Use the lesson well enough to get AI output that is easier to paste into the next workflow

Full Transcript

You can ask the A.I. for exactly the structure you need: a table, a bullet list, JSON, or an email — instead of letting it default to a paragraph. Left to its own devices, the A.I. defaults to a wall of paragraph text.

If you don't ask for a structure, you won't get one. The fix is simple: name the shape you want. Say table, bullet list, JSON object, or email format — the model follows structural instructions reliably.

Structure is a shortcut you give the model, not a constraint you put on it. A clear format request saves everyone rework. Here's a request with no format specified. Tell me about our Basic, Pro, and Enterprise pricing plans.

You'll likely get three dense paragraphs. Now add a format instruction. Compare the plans as a table, with columns for price, users, and key features. Same question, structured output. Naming the format up front means far less rework — no copying a paragraph into a spreadsheet by hand after the fact.

Three formats come up constantly: a table for comparisons, a bullet list for scannable steps, and JSON when you're feeding the output into another tool. To specify a format well: name it explicitly, describe the fields or columns you want, give a short example if it's unusual, and ask it to skip extra commentary.

Without a format request, you get a dense paragraph you have to manually restructure. Ask for the format, and it arrives ready to use. The common mistake is assuming the model knows your exact format. Spell out the structure you expect, especially for JSON with specific fields. Ask for the shape you need, not just the answer. Name the format, describe the fields, give an example when needed, and save yourself the rework.

Next time you need structured information, try adding 'give this to me as a table' or 'as a JSON object' — and watch the format follow instantly.