One of my most popular posts was about writing PRDs (Product Requirement Documents). The first covered overall guidelines for writing specs and ended with a helpful template (here). In the second, I chose a newly released WhatsApp feature and wrote its spec using that template. You can find both posts here and here.

After founding Effective Product Management, I even created a full course on writing PRDs, available here.

Since then, the AI revolution has reshaped many aspects of product management, including how we write specs. It’s now time to explore how AI can accelerate your PRD writing.

First Principles – How I Write Specs

I prefer the term ‘spec’ instead of ‘PRD,’ so let’s stick with that from now on. 

Before jumping into how AI can assist, let’s recap the essential approach for writing great specs:

1. First, master the ‘Why’. You can’t separate the decision to build a feature from the process that convinced you it’s needed. This involves aligning the feature with your product strategy (using frameworks like the CAF Product Framework), responding to urgent requests from strategic customers, or ensuring essential product stability. Whatever your reason, fully understand and confidently stand behind it, including predicting the impact of the feature beforehand.

2. Next, outline the solution (‘What’). Once the problem and impact are clear, outline a high-level solution. Sometimes this is trivial (e.g., adding a “Download as CSV” option). Other times, like optimizing step #3 in your Shopify conversion funnel, the solution might be less obvious, requiring careful consideration.

3. Then, write the spec. As detailed in my earlier posts, I divide specs into four clear parts: motivation and impact, user stories, functional requirements, and analytics.

Turbo-charge Each Step with AI

Here’s how AI can significantly speed up each of these stages:

Mastering the ‘Why’: Use AI tools like Perplexity, ChatGPT, or Claude to enhance your research and discovery. These can help gather market insights, competitor analysis, or even analyze user pain points from various data sources such as call transcripts or product logs. Common use cases (covered in-depth here) include identifying frequent user problems and formalizing their potential impacts.

Outlining the Solution: AI supports solution design by brainstorming solutions, learning best practices from competitors, and quickly generating UI wireframes or prototypes. Claude is probably the best tool for that (as of the time of writing this post). You can even take it a step further, and create interactive prototypes using Claude integrated directly into tools like Figma, gathering valuable feedback before any coding begins. This is one of the product areas that have been through a total revolution due to AI, and you definitely want to leverage it.

Writing the Spec: Use custom GPTs based on your preferred template (or mine here) to automatically draft a structured spec. AI can also help iterate and enrich each section, suggest analytical questions, and proofread your document for clarity before sharing with your team. The way I would do it is that after I’m done with one section, I’d ask it to reiterate and adjust the next sections based on the new knowledge I provided (as it’s hard for the AI to nail it on the first attempt [well… at least for now]).

Used this way, AI accelerates your spec-writing dramatically.

Summary

My original posts established a solid foundation for writing great specs – and those fundamentals still apply. But now, AI advancements enable you to expedite much of the process, significantly streamlining your workflow. This can be especially valuable if writing these documents takes up too much of your time.

Yes, AI can’t yet nail every functional requirement or perfectly break down features into user stories – but if it could, your company might not need you. So celebrate that your brilliance is still essential. 🙂

Next Steps

Try out my GPT, or build your own. Leverage my template, or share your favorite AI-powered spec-writing hacks below!

Good luck!

 —
If you liked this post – feel free to ‘like it’ and share it using the share options below.