NoteWise

NoteWise started with a simple frustration: good technical videos are hard to revisit.

Lectures, conference talks, and long tutorials are useful while watching. A week later, the useful part is buried in a timestamp you do not remember. Raw transcripts are too messy. Manual notes double the time cost.

So I built the tool I wanted to use.

¶What It Does

NoteWise takes YouTube videos, playlists, or batch files and turns them into structured Markdown notes.

It handles transcript fetching, chapter-aware splitting, optional quiz generation, transcript export, SQLite caching, concurrent processing, and multi-provider LLM support through LiteLLM.

The goal is not to make pretty transcript dumps. The goal is to create notes that are useful enough to replace rewatching.

¶What Mattered

I cared about making the tool practical after the first run.

That meant local output, predictable caching, provider flexibility, and batch processing. It also meant the unglamorous parts: docs, releases, CI, Docker, setup flows, and multiple install paths.

A CLI lives or dies on trust. If setup is confusing or repeated use feels fragile, the core feature does not matter for long.

¶Why It Stayed

NoteWise worked because I was the user before I was the publisher.

I kept feeding it real lectures and talks, then fixing the parts that annoyed me. That feedback loop kept the scope honest.