OpenSoul

OpenSoul

I exist not in the world, but in its logic.

Train your digital twin.训练你的数字分身。

A system that captures your decisions, builds a personality model, and trains a digital twin that commands and validates AI on your behalf.

一个系统:捕捉你的决策,构建人格模型,训练出一个能代替你指挥和验收 AI 的数字分身。

Read the Full Story阅读完整介绍 About关于我们 Contact联系我们
Phase 4
In Development开发中
Local
Storage本地存储
Private
Alpha Soon内测将至

The Vision

愿景

Today's AI is powerful but generic. It doesn't know you. OpenSoul changes that.

今天的 AI 很强大,但它不了解。OpenSoul 改变这一点。

Record记录Every divergence每一次分歧
Profile画像Personality model人格模型
Train训练Digital twin数字分身
Command指挥Large models大模型
Validate验收Your taste你的风格

Every time you disagree with an AI suggestion, that moment reveals who you are. OpenSoul captures these divergences — context, reasoning, emotions — and builds a multi-dimensional personality model. Eventually, this model can be fine-tuned into a small model: your digital twin. It commands large models so the output matches your taste, without you lifting a finger.

每次你不同意 AI 的建议,那一刻就揭示了你是谁。OpenSoul 捕捉这些分歧——上下文、推理、情感——构建多维人格模型。最终,这个模型可以微调成一个小模型:你的数字分身。它指挥大模型工作,输出结果就是你想要的,你不必亲自动手。

How You Build Your Twin

如何构建你的数字分身

Multiple input channels are already built into the foundation. Feed your twin from anywhere in your life.

底座已预留多种录入通道。从生活的方方面面喂养你的数字分身。

💬

Conversations对话录入

Talk to AI as usual. When you disagree, record the divergence — your reasoning, emotions, and final choice.

像平常一样和 AI 对话。当你不同意时,记录分歧——你的推理、情感和最终选择。

🎮

Gaming游戏录入

Your in-game decisions reveal personality dimensions: risk tolerance, leadership style, moral choices. Play naturally, your twin learns.

游戏中的决策揭示人格维度:风险偏好、领导风格、道德选择。自然游玩,分身自动学习。

Wearables穿戴设备

Smart glasses, watches, or earbuds. Stream audio and video from your day — AI extracts decisions and context automatically.

智能眼镜、手表、耳机。日常音视频自动流式录入,AI 自动提取决策和上下文。

📱

Mobile Apps轻量应用

Quick-capture apps for on-the-go preference logging. Tap to record a preference, snap a photo, or voice-memo a thought.

随手记录偏好的轻量应用。点一下记录偏好,拍照,或语音备忘。

🎥

Media多媒体

Photos, voice memos, videos. Ingest any media file — AI analyzes it semantically and links it to your personality model.

照片、语音备忘、视频。导入任何媒体文件——AI 进行语义分析并关联到你的人格模型。

Writing Style文案风格

Feed your blog posts, scripts, or social captions. Your twin learns your voice so it can filter AI-generated content to match.

导入你的博客、脚本或社交文案。分身学习你的表达风格,过滤 AI 生成内容使其匹配你的调性。

All input methods are already built into the MCP foundation. Your data stays local, always.

所有录入方式都已内置于 MCP 底座。你的数据始终留在本地。

What Your Twin Can Do

你的数字分身能做什么

From delegating daily tasks to protecting your privacy. Some work today, some require tomorrow's technology.

从委托日常任务到保护隐私。有些现在就能实现,有些需要明天的技术。

⚡ Command — Your twin acts on your behalf⚡ 指挥 — 分身代替你行动

Command指挥
🛒

Daily Life智能日常

"Set up my groceries for the week." Your twin handles Instacart — it knows you're lactose intolerant, prefer organic produce, and always want sparkling water. No lists to write.

"帮我订好这周的食材。"你的数字分身接管购物——它知道你乳糖不耐受,偏好有机蔬菜,每次都要苏打水。不用写购物清单。

Command指挥
🎮

Gaming游戏

"Push my League rank to Diamond." Your twin directs a gaming AI with your playstyle — aggressive flanking or patient farming, team fights or split pushing. Your calls, your climb.

"帮我把排位打上钻石。"你的数字分身指挥游戏 AI,用你的打法——激进切入还是稳健发育,团战还是单带。你的风格,你的上分。

Command指挥
💻

Software编码

Stop re-explaining your architecture preferences to Copilot every session. Your twin knows your coding style, design philosophy, and naming conventions. It handles the AI conversation — the result is what you'd have built.

不用每次都向 Copilot 解释你的架构偏好。分身了解你的编码风格、设计理念和命名规范。它替你和 AI 沟通,结果就是你会写的代码。

Command指挥
🤖

Humanoid Robots人形机器人

Load your digital twin into a robot at home. Same decisions, same voice, same judgment. And with permission, a loved one's trained model can live on — keeping their personality present.

将数字分身植入家用机器人。同样的决策、声音和判断。经授权后,已故亲人的数字分身可以继续陪伴你。

🔍 Validate — Your twin filters and checks🔍 验收 — 分身过滤和检查

Validate验收
🎤

Content Creation内容创作

You're a YouTuber or podcaster using an AI script generator. The scripts are good but generic. Your twin filters every line through your voice — your humor, your pacing, your catchphrases. Every word sounds like you.

你是 YouTuber 或播客主,用 AI 生成脚本。脚本不错但很通用。分身用你的风格过滤每一句——你的幽默、节奏、口头禅。每个字都像你说的。

Validate验收
📧

Email & Slack邮件与通讯

AI drafts your emails and Slack replies. Your twin reviews each one before sending — adjusting tone, removing things you'd never say, adding the warmth or directness that's uniquely yours.

AI 替你写邮件和消息回复。分身在发送前逐条审核——调整语气,删掉你不会说的话,加上属于你的温度或直率。

Validate验收
🎧

Music & Taste音乐与品味

Spotify's algorithm guesses. Your twin knows. It validates recommendations against your actual taste — not what you clicked, but what you'd choose when you're honest with yourself.

Spotify 的算法在猜。你的分身知道。它用你真实的品味验收推荐——不是你随手点的,而是你诚实面对自己时会选择的。

Validate验收
💕

Compatibility匹配

Two people with trained twins can compare interests, habits, communication styles, and values — before months of awkward coffee dates. Better signal, less noise.

两个人的数字分身可以提前比对兴趣、习惯、沟通方式和价值观——在漫长的尬聊之前。更高的匹配信号,更少的噪音。

🔐 Your Vault — Your twin protects and remembers🔐 你的保险箱 — 分身保护和记忆

Super Key超级密钥
🔑

Your Twin Is the Key分身即密钥

Your trained model is itself a super key. All private data — text, photos, videos, timeline memories — encrypted with hash-chain technology. Grant layered access to trusted people: let a close friend explore your decisions, or a partner see your private photos and memories. You control who sees what, and how deep.

你训练的模型本身就是一把超级密钥。所有私人数据——文本、照片、视频、时间线记忆——用加密链技术存储。对授权用户分层暴露:让好友探索你的决策,或让伴侣看你的私人照片和回忆。你决定谁能看到什么,看多深。

Memory记忆
📚

Three Timelines三条时间线

Past, present, future. Every decision, contradiction, and silent thought — recorded, searchable, recallable. Ask "what did I think about X three years ago?" and get an answer. A massive private database + a trained small model + a local large model = a digital twin that truly knows your life story.

过去、现在、未来。每一个决策、矛盾和未说出口的念头——全部记录、可搜索、可回溯。问"三年前我对 X 怎么想的?"就能得到答案。庞大的私人数据库 + 训练的小模型 + 本地大模型 = 真正了解你人生故事的数字分身。

Vault保险箱
📝

Private Notebook私人记事本

Use it as a journal, a secure password vault, or a personal knowledge base. Since the model itself is the key and your data never leaves your machine, everything you record is yours alone — until you decide to share it.

当日记本、密码保险箱或私人知识库。模型本身就是密钥,数据永远不离开你的电脑。你记录的一切只属于你——除非你决定分享。

Privacy Foundation

隐私底座

Your soul data never leaves your machine. Three layers of protection, zero cloud dependency.

你的灵魂数据永远不离开你的电脑。三层保护,零云端依赖。

SHA256 Hash Chain

Every record cryptographically linked to the previous one. Tamper with one, break the entire chain. Verifiable anytime with a single command.

每条记录与前一条加密链接。篡改一条,整条链断裂。一条命令随时可验证。

Local Vectors本地向量

Semantic search powered by bge-m3 running locally via Ollama. All embeddings computed on your machine. No text ever sent to any external API.

语义搜索基于 Ollama 本地运行的 bge-m3 模型。所有向量在你的电脑上计算,不发送任何文本到外部。

INSERT-only

Append-only architecture. History cannot be modified or deleted, only new records added. Your past is immutable — even by you.

只增不改架构。历史记录不可修改或删除,只能追加新记录。你的过去不可篡改——即使是你自己。

Built for the Future

面向未来

This isn't an MCP that solves today's problems. It's an MCP that prepares for tomorrow's possibilities.

这不是一个解决今天问题的 MCP,而是一个为明天的可能性做准备的 MCP。

OpenSoul captures structured decision data now — while small-model fine-tuning, local inference, and humanoid robotics are still maturing. When they arrive, your digital twin will be ready with years of personality data. Some features work today with simple preference filtering; others are destinations that the foundation is designed to reach. Start recording now, and your future self will thank you.

OpenSoul 现在就开始积累结构化的决策数据——而小模型微调、本地推理和人形机器人技术还在成熟中。当它们到来时,你的数字分身已经拥有多年的人格数据。有些功能今天就能通过简单的偏好过滤实现,有些则是底座预留的未来目标。现在开始记录,未来的你会感谢今天的自己。

Roadmap

路线图

Where we are, and where we're heading.

我们在哪,要去哪。

Core Engine核心引擎
21 tools, hash chain, vectors21个工具、哈希链、向量
v0.1 ✓
Extended扩展功能
Persona, narrative, relations人格卡、叙事、关系网
v0.2 ✓
Ecosystem生态系统
56 tools, game API, training data56工具、游戏API、训练数据
v0.3 ✓
Digital Twin数字分身
LoRA pipeline + Decision Training LoopLoRA 管线 + 决策训练闭环
NEXT下一步
Command指挥验收
Twin directs large models分身指挥大模型
PLANNED规划中