THOUGHTER · The AI inside your notes

The AI in your notes,
a thinking assistant.

When you ask a question, it searches across your own past notes first and quotes your words back to you instead of giving generic advice.

Free & open source · Windows / Android · Thoughter is experimental and uses your own AI provider

Thoughter conversation screenshot: after reading two months of notes, it points out the user keeps looking for empty space
Real in-app screenshot (Chinese UI)

It reads your notes before it speaks.

It looks for threads in what you wrote weeks or months ago, and quotes your own words back to you instead of giving generic advice.

  • Searches across all your notes, not just the one in front of you
  • You can expand each answer to see which notes it read

I keep thinking about “slowing down.” Have I written about this before?

Reading your notes…Read 3 notes

  1. A patch of plaster fell off a wall in the old town, showing older red brick underneath.

  2. Foggy morning. The steam from the bun reached my hands before the sun did.

  3. In the cello suite: the stillness right after a note breaks off.

You never use the words “slow down,” but you keep recording the feeling: the peeling brick, steam in the fog, the stillness between notes. That is how you slow down.

Tell it once. It remembers.

When you mention a habit — say, “keep answers short” — it writes it down, so you never have to repeat yourself.

If you turn it on, it also reads your recent notes in the background, sums up your writing habits and tastes, and shows how many notes each conclusion is based on.

  • Memory lives only on your device
  • Every entry can be viewed, edited or deleted
Memory page: one entry from chat, one distilled from notes

Your words, your call.

When you ask for a polish, changes arrive as a suggestion card. Until you tap “Apply,” not a single character of your original changes.

  • See exactly what changed before you decide
  • Open it in the editor if you want to tweak it yourself
EditPolish: refactoring notes

This afternoon I was basically goingwent through how the project modules call each other, and honestly the overall logic feels way too long and too tangled. A lot of codeCode that could have been merged got split into several layers, justlayers because my thinking back then was really scattered. Before refactoring, I really need to get the main thread clear first, so I don’t end up goinggo in circles again.

✓ Applied to note Open in editor

Your notes stay on your device.

ThoughtEcho is a local-first notes app. Thoughter only works when you ask, and only takes the few passages it needs.

No server of ours

ThoughtEcho runs no relay server, collects none of your notes and needs no account.

Only relevant passages

When you ask, only the few related passages go straight to the model provider you chose.

Fully offline, if you like

Connect a local model via Ollama or LM Studio and talk to Thoughter with no internet at all.

Works with DeepSeek · OpenAI · Qwen · Kimi · SiliconFlow · Ollama · LM Studio · any OpenAI-compatible API

Questions

Does Thoughter upload all my notes?

No. Your notes stay on your device. Only when you ask Thoughter something does ThoughtEcho send the related passages over HTTPS directly to the provider you configured — no ThoughtEcho server in between.

Does Thoughter cost anything?

ThoughtEcho is free and open source, with no ads or in-app purchases. Thoughter uses your own API key, billed by your provider per use. Local models like Ollama cost nothing.

Can I use ThoughtEcho without AI?

Yes. Writing, rich-text editing, tags and WebDAV sync all work without AI. Thoughter is optional.

Can I delete what it remembers?

Yes. On the Thoughter memory page you can view, edit or delete every entry, or switch memory off entirely in AI settings.

Write it down. It’ll remember.

ThoughtEcho is free, open source and ad-free.