The problem: you read something great, but your notes are a mess
You finish a 3,000-word article. You want to save the key ideas into Obsidian, Logseq, or Notion. But the article does not have clean section breaks - it is one long wall of text.
So you either:
- Copy-paste the whole thing into one giant note (useless - you will never find anything)
- Manually highlight and split it into sections (takes 10-15 minutes per article)
- Pay for an AI summarizer that ships your text to a server (and starts building a profile on you)
None of these are good. There is a better way.
The solution: split by meaning, not by word count
TopicSplit is a free, offline tool that reads any pasted text and finds where the meaning shifts - then cuts there. It is not a summarizer. It is a semantic segmenter.
You paste a messy article, transcript, or meeting notes. It outputs clean markdown with ### Topic 1, ### Topic 2, etc. Each segment is a self-contained chunk of meaning.
Try it now - free, no install, no tracking
Paste any text. Get clean topic segments in one click.
Open TopicSplit →How it works (30 seconds)
- Paste any article, transcript, or notes into the left panel.
- Drag the Sensitivity slider - higher = more segments, lower = fewer.
- Click "Split into topics" - get clean markdown on the right.
- Copy or download - straight into Obsidian, Logseq, or Notion.
Why this beats the alternatives
Manual splitting
- Takes 10-15 min per article
- Inconsistent - depends on your energy
- Easy to miss key transitions
- Does not scale
TopicSplit
- Takes 5 seconds
- Same quality every time
- Finds shifts you would miss
- Works on any length of text
Built for PKM workflows
If you practice Zettelkasten, atomic note-taking, or any Personal Knowledge Management system, TopicSplit fits naturally:
- Obsidian users: Download the markdown directly into your vault
- Logseq users: Split journal entries or research into topic blocks
- Notion users: Turn meeting notes into organized sections
- Researchers: Segment interview transcripts by theme
- Writers: Structure long-form content into clear sections
100% offline - your data never leaves your browser
TopicSplit runs entirely client-side. There is no server, no API key, no tracking. Your text is processed locally and never transmitted. This is not a privacy policy - it is an architectural guarantee.
Free forever, open source
TopicSplit is MIT-licensed. You can use it, modify it, share it, or build on it. No paywalls, no feature tiers, no "pro version."
If it saved you time, a $1 tip keeps the next tool coming:
Support free, private tools
Every dollar goes back into building more offline, tracking-free tools.
☕ Tip $1 on Ko-fi 💳 Tip $1 on PayPalFAQ
Does it work in languages other than English?
Yes - the segmentation algorithm is language-agnostic. It works on any text where topic shifts are marked by vocabulary changes.
What is the longest text I can paste?
There is no hard limit. It has been tested on transcripts up to 50,000 words. Performance depends on your browser.
Do I need to install anything?
No. Open the live demo in any browser. Or download index.html and open it locally - works offline.
Can I use this for commercial work?
Yes. MIT license means you can use it for anything - personal, commercial, whatever.
Is there an API?
No API. That is the point. If you want to integrate it into a pipeline, download the source and run it locally.