[item007]

In Concerns the “action” isn’t generic, it’s anchored to a target/project you set first.

The flow I’m exploring is: you define a small number of active targets (e.g. “ship feature X”, “prepare for interview Y”). Then when you save/read something, the system searches your existing library (notes/links/email/posts/etc.) against that target and suggests a few candidate next steps or plans that are specifically useful for that target. You pick one (or dismiss them), so it’s more “menu of options” than “AI tells you what to do”.

Example 1 (technical): target = “build a small Kotlin app”. From a Kotlin article + your saved repos, it might suggest: “start with template A”, “try library B for state management”, or “do a 30-min spike to validate architecture C”.

Example 2 (research/learning): target = “write a short brief on topic Z”. From your saved posts, it might propose: “3 key claims + 2 counterpoints”, plus a short outline you can accept/edit.

So “action” = a target-linked next step or plan proposal, chosen by you — not turning every summary into a task.

[koakuma-chan]

I feel like you could just use AI for that without managing your own library (plus it got pretty good at web searches). At the very least, I probably don't collect enough articles for that to work. Maybe if there is like a shared managed collection.

[item007]

I think we might be talking past each other a bit. I’m not trying to build “a better personal library manager” or “AI search over your notes”. The core idea is goal-first: you set a small number of active targets/projects, and the system uses whatever sources you already have (even a small amount) to propose concrete options you can choose from: a plan, a checklist, tradeoffs, or a next experiment that moves that target forward. The value isn’t “organizing”, it’s reducing the thinking overhead of turning scattered inputs into an executable next step.

And you’re right that many people don’t collect enough personally — that’s why I’m also considering a hybrid where your own saves provide personalization, but a shared/managed collection (or public sources) fills the gaps.

In your case, would you find this useful if the output was “one good plan/next step per target” even when your personal saves are sparse, or do you prefer it to be entirely web-driven unless you opt in?

[koakuma-chan]

What I'm saying is that, if I need to know how to achieve X, I can go into, say, ChatGPT, and ask, "how to achieve X?"—I don't see why I would need a separate app for that. I don't see much value in an app that just inserts some notes or blog posts into the LLM's context for me.

But it could just be that I don't collect that kind of notes or something. Maybe this would work for somebody else.

[item007]

Understand and thanks again.

[koakuma-chan]

No problem