AI WRITING COMPANION

翻阅旧日片羽，聊聊当下所想。

An AI companion living inside your local notes. It understands your voice, explores past thoughts, presents structured proposal cards, and accompanies your writing with private local memory.

Download ThoughtEcho Explore Philosophy

Local-First · Full Ownership

Proposal Cards · You Decide

Personal Memory · Learns Your Voice

Open API · Offline Model Support

THREE CORE PRINCIPLES

Think, Understand, and Respect Your Voice

Not a generic chatbot detached from reality, but a thoughtful companion rooted deeply in your personal notes.

01 · ACTIVE EXPLORER

The Mind That Explores

Beyond answering prompts — actively explores your personal universe of notes.

When you ponder an idea, Thoughter calls native tools to browse your local notes from months ago — weaving forgotten fragments into today's creative breakthrough.

Smart retrieval across local notes — connects past thoughts with today
Native tool calling — verifies facts instead of making things up

Spark · Today's Reflection

"On a cloudy dusk in Hangzhou, what unspoken thoughts are resting in your heart with the evening breeze?"

Real native UI: Executed 4 actions, surfaced Jobs quote for deep insight

02 · PERSONAL MEMORY

The Companion Who Remembers

Learns your voice and habits across chats — more intuitive with every note.

No need to re-introduce yourself every time. It observes your writing voice and reading taste to build personal local profiles. Kept 100% on your device, clearable and editable anytime.

Personal profile layers (voice & taste) adapt to your style
Isolated local database (agent_memory.db) — zero cloud tracking

Real memory dashboard: Collector voice and classical/minimalist taste

03 · CREATIVE SOVEREIGNTY

A Respectful Partner

Offers thoughtful proposals, but the pen stays firmly in your hand.

Unlike tools that overwrite your writing, Thoughter respects your creative boundary. Revisions arrive as independent Proposal Cards. Accept with one click or fine-tune in the editor — your original draft is never touched.

Independent proposal cards — never overwrites your writing without consent
Quick edit and full editor integration — you hold total editorial control

Real Proposal Card: provides instant [Open in Editor] and [Apply Revision] actions

DESIGN DIFFERENCES

Key Design Differences

An objective look at how Thoughter differs from generic chat bots and cloud notes.

Feature / Dimension	ThoughtEcho Thoughter	Generic Web AI Chat	Traditional Cloud Note AI
Notebook Integration	Deeply explores and retrieves relevant notes from your notebook	Disconnected context; requires manual copy-pasting	Relies on centralized cloud-side index search
Suggestion Presentation	Independent proposal cards; accept directly or fine-tune in editor	Plain text output inside the chat stream only	Often directly overwrites original notes
Preference & Memory	Learns your voice across chats; clearable and manageable anytime	Prompts and profiles stored on vendor cloud	Deeply locked into proprietary cloud ecosystem
Model Connectivity	Standard OpenAI protocol + local offline models via Ollama	Locked to a single provider	Closed models with no BYOK or provider choice

MODEL COMPATIBILITY

Open & Flexible Model Support

ThoughtEcho adopts standard AI communication protocols. Connect to mainstream providers like DeepSeek or OpenAI, or run completely offline using Ollama or LM Studio.

DeepSeek

OpenAI

Ollama (Local Offline)

LM Studio

SiliconFlow

Qwen

Moonshot Kimi

Custom OpenAI Endpoint

QUESTIONS & ANSWERS

Frequently Asked Questions

Clear, honest answers about privacy, billing, and offline use.

Does Thoughter upload all my notes to AI servers? +

No. Your notes are stored securely on your own device by default. Only when you ask Thoughter a question or request a note summary does ThoughtEcho send relevant context snippets directly to your configured provider over HTTPS. ThoughtEcho operates no intermediate server and collects no note data.

Is Thoughter free to use? +

ThoughtEcho itself is completely free and open-source with no ads or subscriptions. Using Thoughter requires your own API key (e.g. DeepSeek or OpenAI), billed directly by the provider based on token usage. Running local models via Ollama or LM Studio is entirely free.

Why doesn't Thoughter overwrite notes directly? +

This is a core design principle: "Creative sovereignty always belongs to you." All creations and revisions arrive as structured proposal cards that require your review and confirmation before applying to your notebook.

How does the memory system work? +

Thoughter maintains a dedicated memory profile for your writing style and phrasing preferences across chats. You can ask Thoughter to forget any detail or clear all memories anytime in Settings › AI Settings.

Can I use Thoughter offline without internet? +

Yes. ThoughtEcho supports local model endpoints like Ollama and LM Studio. With a local model running on your machine, you can chat with Thoughter and explore your notes completely offline.