Great piece, Bandan! I’ve written every kind of PRD, from massive, 400+ page Waterfall “Project Requirements Docs” to Amazon-style PR/FAQs, to Notion docs, to one-pagers. The key thread? There’s a small set of strategic context choices that need to be articulated for both humans and LLMs for them to be able to collaborate effectively.
The high value butsharp customer insight, problem framing and business context are almost always things I don't leave to AI. But it can often help in filling in my thinking gaps.
At-least the PMs I have worked with never thought documentation length matters anyways, but creating alignment is a whole different skillset that needs to be built even when you're writing shorter more concise documents
Interesting! In my last product we maintained a pretty large Confluence for requirements documentation, but it was never quite kept up with every change and experiment to really reflect what the product was or why changes were made. I wonder if that could be the gap AI could fill in PRDs. I believe the initial crafting should come from real, informed thoughts about your business and customers (not "generic industry patterns" as you say) and AI could come in to help update documentation with decisions from meeting notes, emails, slack conversations, etc. That would actually save all of us time and could create better artifacts! I wonder what tools would be good to make this work.
Hey Leda, thanks for sharing your thoughts. There is no one rule on when to use or not use AI in your documentation. But upfront high value customer and business decisions are something a PM should directly analyze and refer to in PRD (in form of insights, problem statement framing, asking the right Why questions), leaving the structuring and 'filling in the gaps' to AI.
Great piece, Bandan! I’ve written every kind of PRD, from massive, 400+ page Waterfall “Project Requirements Docs” to Amazon-style PR/FAQs, to Notion docs, to one-pagers. The key thread? There’s a small set of strategic context choices that need to be articulated for both humans and LLMs for them to be able to collaborate effectively.
The high value butsharp customer insight, problem framing and business context are almost always things I don't leave to AI. But it can often help in filling in my thinking gaps.
Yes! I’ve been experimenting with strategy prompt engineering for a while, and paid subscribers get my two tested prompts in this week’s newsletter.
The key?
You have to provide context, and the LLM helps ask the tough, probing questions to test your strategy and deepen your strategic thinking skills.
That was good! Great article, thanks for sharing. Clear, well-structured, and a refreshing take on why it still matters.
I am glad you liked it William
Alignment usually matters more than documentation length ever will.
At-least the PMs I have worked with never thought documentation length matters anyways, but creating alignment is a whole different skillset that needs to be built even when you're writing shorter more concise documents
Speed without shared understanding just gets you to the wrong place faster.
PRD as shared understanding is a great way to frame it.
Interesting! In my last product we maintained a pretty large Confluence for requirements documentation, but it was never quite kept up with every change and experiment to really reflect what the product was or why changes were made. I wonder if that could be the gap AI could fill in PRDs. I believe the initial crafting should come from real, informed thoughts about your business and customers (not "generic industry patterns" as you say) and AI could come in to help update documentation with decisions from meeting notes, emails, slack conversations, etc. That would actually save all of us time and could create better artifacts! I wonder what tools would be good to make this work.
Hey Leda, thanks for sharing your thoughts. There is no one rule on when to use or not use AI in your documentation. But upfront high value customer and business decisions are something a PM should directly analyze and refer to in PRD (in form of insights, problem statement framing, asking the right Why questions), leaving the structuring and 'filling in the gaps' to AI.