Start here · 3:26
The first draft is not the answer
This lesson explains what a prompt framework is, then shows why the first AI draft is only raw material and how a self-critique pass improves it.
- A prompt framework is a reusable checklist of the parts a good prompt contains, usually named with an acronym. It helps because the AI fills in missing parts with generic guesses.
- Frameworks make prompts repeatable. You fill the same parts each time, save the prompt as a template and share it with colleagues. They are not magic words. The value comes from the detail you put in.
- Self-critique is a prompting technique from AI research. Instead of slots to fill, it changes how the AI works through a task. You can add it on top of any fill-in framework.
- Models are better critics than first-draft writers. A review pass catches much of what the first pass missed. The first draft is raw material, not the answer.
- Self-critique techniques are one of six families in The Prompt Report survey of prompting. Anthropic calls draft, review and refine the most common prompt chaining pattern.
- There are well over sixty named frameworks, and most remix the same ingredients: role, task, context, audience, format, constraints and examples.
| Term | What it means |
|---|---|
| Prompt | The request you type into an AI tool |
| Prompt framework | A reusable checklist for that request |
| Self-critique | Asking the AI to review and improve its own work |
| Draft, review, refine | The most common prompt chaining pattern |
Do this in your own tools, step by step. Ticks are saved on this device.
What is a prompt framework?
A reusable checklist of the parts a good prompt contains, usually named with an acronym.
Why is the first draft not the answer?
Models are better critics than first-draft writers, so a review pass catches much of what the first pass missed.
What is the most common prompt chaining pattern according to Anthropic?
Draft, review and refine.
Adam Welcome to Self-Critique: Self-Refine, Chain of Verification and Pre-Mortems. This course is part of Complete AI Training; you'll find the full library at complete A I training dot com. It takes six short chapters.
Paula This part explains what a prompt framework is, then shows why the first draft from an AI is only a starting point. You will learn how a review pass catches what the first pass missed.
Adam A prompt is the request you type into an AI tool such as ChatGPT, Claude or Gemini. A prompt framework is a reusable structure for that request. Think of it as a short checklist of the parts a good prompt contains, usually named with an acronym so it is easy to remember.
Paula Why does a framework help? The AI fills in whatever you leave out with generic guesses. A framework makes sure you include the parts that change the answer, such as who it is for, what you want to achieve and the format you need. The detail you add is what makes the difference.
Adam Frameworks also make prompts repeatable. You fill the same parts each time, save the prompt as a template and share it with colleagues. They are not magic words. The value comes from the detail you put in each part. They help most when a request is complex or open to interpretation. A quick, simple question can stay short.
Paula There are well over sixty named frameworks, and most of them remix the same ingredients: role, task, context, audience, format, constraints and examples. This course covers a different kind of technique. Self-critique is a prompting technique from AI research. Instead of slots to fill, it changes how the AI works through a task, and you can add it on top of any fill-in framework.
Adam Now to the main idea. Models are better critics than first-draft writers. A review pass catches much of what the first pass missed. The first draft is not the answer. It is raw material. Your job is to ask the AI to look at its own work and improve it.
Paula Self-critique techniques are one of six families in The Prompt Report survey of prompting. That survey grouped the many prompting methods into six families. Self-critique is one of them. It is not a fringe trick. It sits alongside the other major ways researchers have found to get better results from AI.
Adam Anthropic calls draft, review and refine the most common prompt chaining pattern. That means you ask for a draft, then ask the AI to review it, then ask it to refine based on that review. It is a simple loop you can run in any chat tool. Let us see what that looks like in a real job.
Paula Here is the anatomy of a self-critique prompt for a writer working on a blog post draft. Notice the slots: the task, the audience, the draft itself, the critique request and the revision instruction. Each slot gives the AI something specific to work with.
Adam Now watch it in action. The writer pastes the draft and asks for a critique, then a revision. The AI's reply starts by naming the weaknesses. That is the review pass doing its job. The first draft was not the answer. The critique made it better.
Paula The takeaway is simple. Do not accept the first draft. Ask the AI to critique its own work, then revise. That one extra step is the difference between a rough start and something you can actually use. In the next part, we will go deeper into the self-refine loop.
Adam Your turn. Take a recent request you made to an AI tool. Paste the AI's first answer back in and ask it to list three weaknesses, then rewrite. Compare the two versions.
Get the other 5 chapters, the exam and your certificate for LinkedIn, plus all 700+ AI courses.