Manual Setup for Product Management with AI(and looking for feedback)
Over the past 12 months I moved almost all my PM work into GitHub repos and Markdown files. Discovery, scoping, prototyping - running through a chat window or terminal. The unlock was maintaining a structured context folder per product: codebase, interviews, analytics, docs - all in one place, connected to live sources.
When that context is current, every workflow (scoping a feature, synthesizing interviews, drafting a spec) runs fast and produces output I can actually trust.
I'm now building a tool that structures and automates exactly this — building the product context and executing PM workflows on top of it.
Would love feedback, opinions, pushback. Is this a workflow you recognize? What would make you actually use something like this?
resbeefspat The cold start problem you mentioned is the hardest part honestly. I spent about 3 weeks just figuring out how to keep the context folder from, going stale before I wired up some webhook triggers to auto-sync it whenever source data changed. Once that was running the actual PM workflows got way more reliable because I wasn't second-guessing whether the context was current. Mar 23 2 likes
Such_Grace The context-going-stale problem is what kills this workflow for most people, you build the perfect structure and then it drifts within a week. I ended up using Latenode to wire webhook triggers that auto-sync whenever source data changes, which mostly solved the manual upkeep that was eating my time. Mar 23 2 likes