From a year-old research doc to a PRD, with Claude
Handing a year of research notes to Claude and getting back a PRD: the plan I keep coming back to when the project gets fuzzy.
Turn my 2025 research into a product plan I could build from
When I get lost, the PRD is my map back.
By spring 2026, my Check In App research was more than a year old. The insights were solid, but they lived in scattered notes. Before designing anything, I wanted one document that said what I was building and why.
So I gave Claude my research in Cowork and asked it to write a product requirements document (PRD), saved straight into my Obsidian vault. I reviewed it, made changes and approved it. Here's what it covers:
- The problem, in three parts. Redundant questions, questions with no clear purpose, and waiting time that feels wasted.
- Five MVP features. A "Why are we asking this?" note on sensitive fields, questions filtered by reason for visit, pre-filling before you arrive, a quick confirm flow for returning patients, and a waiting room view.
- Eight key screens, including how the app handles an uncomfortable question and what happens when you don't know an answer.
- A V2 list. An AI form companion, a live queue and a medical profile vault, kept out of the first build on purpose.
- Open questions, ranked. How returning patients sign in, how skipping works, and how to structure "reason for visit."
It also named the differentiator: none of the competitors I looked at lead with explaining why each question is asked. That became the heart of the product.
What worked
- The research held up. A year later, the same three problems were still the right ones to solve.
- Scope on paper. Writing down what's out (V2 ideas, clinic dashboards, insurance) kept the build small enough to finish.
- A map for when I'm lost. When the project gets fuzzy, I go back to the PRD to remember what matters.
What didn't
- Only the patient's side. My research never covered the front desk or providers, so the PRD has to assume how their workflows go. That's a gap to close before anything real ships.
- I didn't keep handing it to Claude. When I moved on to styling in May, I worked from memory instead of giving Claude the PRD as context. That's the gap I write about in "Give Claude the why."
When I get lost, the PRD is my map back.
A side project that pauses for a year needs something to restart from. Research tells you what's wrong. A PRD turns that into decisions: what to build first, what to leave out and what's still unknown.
Having Claude draft it meant I spent my time on judgment instead of formatting: reviewing, cutting and deciding. The thinking was mine. The first draft didn't have to be.