Consilience把你的文件，变成 AI 真正能推理的知识图谱。
只管书写、拖入文件，或照常写代码——Consilience 会在你的笔记、代码、设计与地图之间悄悄生长出一张知识图谱。随后一个智能体在其之上工作：提出复杂问题，或把整件任务交给它，它会用上你所构建的一切。
- 免费开始
- 支持 7 种语言
- 在你的设备上运行
在知识图谱之上工作，智能体便截然不同。
Consilience 会自动将你的工作区织成一张本体论知识图谱。于是，同样一句话，也能以普通 AI 智能体触及不到的深度来作答、来办事。
每一份资料都汇成一张知识图谱。
把表格、Word、PDF、图片和网址按原样导入，关键概念与关系便会自动编织成知识图谱。智能体在其之上推理，作答并办事——真正的第二大脑。
- 忠实转换，绝不压成摘要
- 一保存就以概念和关系相连
- 没有文件夹也没有标签，主题社群自行浮现
用 markdown 文档构建你的知识图谱。
一个可自由驾驭表格、公式、图表、维基链接和斜杠菜单的编辑器。改动一则笔记，知识图谱也随之变化。
- 真正的 markdown，随处可开
- 维基链接把笔记与笔记相连
- 笔记始终是你的，作为纯粹的原件
代码智能。
理解代码的结构与含义，精准定位该改之处——继而规划、查找、修复并运行以验证。
- 按代码所做之事查找，而非仅靠文本匹配
- 追踪调用方与导入，看清一次改动的影响
- CodeBridge 把函数背后的设计笔记引入工作
把地点化为洞见的地图画布。
地理编码把图谱中的地址和地点转成坐标，智能体为每次请求创作一张交互地图——热点、按区汇总、可达范围、路线。坐标计算交给程序库，绝不臆测，因此始终精确。
- 地理编码把地址和地点转成坐标
- 热点、区域汇总、可达范围与路线
- 在地图上选一个点，顺着它找到图谱节点
一个智能体，无处不在
不只是笔记。代码、终端、外部应用、网络，乃至屏幕——同一个智能体在你设定的权限下全部搞定。
计划模式
复杂工作先立计划；在你批准前编辑被锁住
多文件编辑
跨多个文件规划改动，一次性应用
终端 · 运行命令
运行构建和测试，读取结果，决定下一步
直接操控电脑
观察屏幕，用鼠标和键盘操作桌面应用
外部应用连接
在你的权限下直接操作 Notion 等外部应用
网页搜索
需要时从网上获取最新信息作为依据
本体搜索
顺着概念与关系，在整个工作空间中找出所需
主题地图
以地图呈现你的笔记聚成了哪些主题
领域词汇
学习你所在领域的术语，按你的说法理解
同义词 · 关系建议
找出相合的含义和隐藏的关联并率先提议
有出处的答案
不是臆测，而是你写下的句子，直到出处笔记
markdown 编辑器
表格、公式和图表，就在真正的 markdown 文件里
多模型选择
为每项任务亲自挑选最新模型和推理强度
经验记忆
把你教过的和它确认过的存为笔记，越用越聪明
Git 集成
就地审阅改动，处理提交与分支
权限控制
预先为每个工具设定运行权限，只在必须时才询问
工作空间共享
把工作空间以文件或链接共享，原样打开
地理编码 · 地图
把图谱中的地点转成坐标放上地图，并在其上分析
与知识共事的全新方式。
各行各业的团队真实遇到的问题，以及他们如何解决。滑动卡片阅读这些案例。
32 pages of law, one question
Construction staff stop paging through the subcontracting law themselves. They ask in their own words, and the answer comes back with the rule it is based on and a link to the original text.
Ask the 32-page Subcontracting Act something like "if we pay more than 60 days late, what about the late-payment interest?" and get an answer with the supporting clause and its source link attached.
Customer PoC · Demo阅读案例 →Personal notesYou don't need the right search term
A product manager with five years and 1,800 notes finds his own writing using a concept that appears nowhere in the title.
Keyword search only finds notes that contain the words you typed. Ask a question instead and you also get the notes that cover the same topic in different words. Ask about the subscription economy and a note titled "Membership retention ideas" comes back, even though it never uses that phrase.
Example · Hypothetical scenario阅读案例 →ManufacturingOne search that finds every word for the same thing
The same machine is called three different things in three systems, so searching one name never returns the history filed under the other two. Confirming the three as one machine is what makes 15 years of scanned logs, 30,000 MES rows, and a weekly interview recording answer as a single history.
The answer is written down, but you only find it if you search the exact word it was written under. Consilience links the different words people use for the same thing, so searching any one of them returns all of it, with a link to the record each answer came from.
Example · Hypothetical scenario阅读案例 →Pharma & Chemical R&DSearch internal documents and outside papers without switching tools
Internal reports, patents, and outside papers all answer in one chat, so there is no toggling between three systems. Every answer carries its source: the document name and page, the patent conditions, the paper metadata.
Ask internal technical reports, patents, and external papers in one chat without moving windows, and get a report where the document name and page, the patent conditions, and the paper metadata are attached as sources.
Customer PoC · Demo阅读案例 →Software teamsWhy did we build it this way?
The CTO of a 30-person B2B SaaS company questions 200-odd meeting notes and a pile of architecture decision records one decision at a time. The original sentence and its revision trail outlast the person who wrote it.
Ask what your team decided on any topic and the decisions come back in order, starting with the most recent. Each one names who made it and links to the exact line in the meeting notes where it was recorded. If a decision was later reversed, it is marked as replaced, so you never act on one that no longer stands.
Example · Hypothetical scenario阅读案例 →Personal investingFollow a stock and get real time news
A retail investor with eight years of research notes pulls the assumption he wrote at buy time back up, sets a watch on the words that assumption hangs on, and gets a notification when new stories come in.
The note you wrote when you bought a stock and any news you clip about it later end up on the same page. When an incoming story mentions something you're watching, you get a "Watch: capacity expansion" notification. Ask about that stock and the answer quotes your note and the article side by side.
Example · Hypothetical scenario阅读案例 →为你解答。
我的数据安全吗？
安全。Consilience 本地优先——你的笔记和知识图谱都在你的设备上。我们不会用你的笔记训练任何模型，索引与搜索在设备上运行，网络访问默认关闭。智能体在行动前会请求许可。
离线能用吗？
应用、笔记和知识图谱都能离线使用。智能体只有在访问语言模型时才需要联网；你构建的一切都留在设备上，照常读取。
可以导入什么？
笔记、PDF、Word、表格、图片、网页乃至代码——按完整内容忠实导入，绝不压成摘要。
能从 Obsidian 或 Notion 迁移吗？
可以。把你的 Markdown 和文件原样带来。你的笔记仍是随处可开、随时可导出的标准文件——没有厂商锁定。
支持哪些平台？
目前是 macOS 与 Windows。免费即可开始，并会自动下载适合你系统的版本。
付费计划改变了什么？
只改变你能用多少智能体。包括免费在内的每个计划都是完整的工作空间；付费计划只是提高智能体用量。随时可取消。