How to Write PRDs with AI

The Paste and Ask Pattern

The most effective way to write with AI is the paste and ask pattern. You paste relevant information into your prompt. Then you ask AI to draft a specific section.

This works because AI needs context. Without context, AI makes assumptions that are often wrong. With context, it can generate relevant drafts.

Structuring Your PRD Prompt

Here is a reliable prompt structure:

1. Paste the background: What is your product? Who are your users? What problem are you solving?
2. Paste the specific input: Feature description, user research summary, or stakeholder feedback.
3. Ask for the output: Draft a problem statement, or draft the goals section, or list success metrics.

Be specific about what you want. Do not ask AI to write the entire PRD at once. Break it into sections.

Sample PRD Prompt

Here is a template you can use:

"Our product is a project management tool for small teams. Our users are team leads and individual contributors who want to track tasks without complexity. We are adding AI-powered task suggestions based on team patterns.

Context: Sales teams have requested automation for recurring tasks like weekly status updates and follow-up scheduling. Current feedback from 15 user interviews shows 80% want AI to suggest when a task should recur, but only 40% want full automation.

Please draft a problem statement and goals section for this feature. The problem statement should explain the current friction. The goals should reflect the findings from user research."

AI will give you a draft. You read it. You edit it. You add your strategic context. You are done.

Writing Different PRD Sections

Different sections need different prompts.

For the problem statement, give AI user pain points and current workarounds. Ask it to draft a one or two paragraph explanation.

For goals, list what you want to achieve and who benefits. Ask AI to make each goal specific and measurable.

For requirements, describe the feature and ask AI to list 5-10 acceptance criteria.

For success metrics, tell AI what you care about: adoption, time saved, user satisfaction. Ask it to suggest 3-5 metrics and how to measure them.

Common Mistakes to Avoid

Mistake one: Writing vague prompts. "Write a PRD for a new feature." This does not work. AI has no context. You get generic output.

Solution: Always paste context. Tell AI about your users, your product, your constraints.

Mistake two: Accepting the first draft as final. AI drafts are starting points, not finished work. You must refine.

Solution: Treat AI output as a draft you will improve. Add your knowledge. Add details AI missed. Remove things that sound wrong.

Mistake three: Using AI for decisions instead of writing. Do not ask AI which feature to build. Ask AI to help you articulate the feature you have already decided to build.

Solution: Make decisions yourself. Use AI to document and refine your thinking.

Iterating on Drafts

After AI gives you a draft, ask follow-up questions. Ask it to make something more detailed. Ask it to simplify jargon. Ask it to reframe something for a specific audience.

Example follow-up: "This draft is too technical for non-engineers. Rewrite the success metrics in plain language that a marketer would understand."

Each iteration takes seconds. You can refine drafts much faster than writing from scratch.