Fabric

AI 驱动的第二大脑工作区,自动整理你的笔记、文件与链接,用语义搜索和 AI 助手随时找回并用好一切

In-depth Report

  • Fabric is an AI-driven workspace positioned as a "second brain". It advocates integrating all notes, files, links, screenshots, voices and emails into the same space that can be organized by itself. It can be searched using natural language and can be asked using AI assistants. Its biggest selling point is not "another note-taking software", but "you don't organize it, it connects it for you" - relying on semantic search and automatic association to turn data scattered in more than a dozen apps into a conversational knowledge base. The product is still in its early stages (App Store version number 0.8.

  • The official domain name of Fabric is fabric.so, and the entrance recorded in daily express is getfabric.io (the latter will jump). The developer registered in the App Store is "Fabric Internet" (corporate entity WFHLAND, Inc.), and the iOS App is called "Fabric: AI Notes, Files, Ideas". The product follows the typical AI knowledge management track and directly competes with Notion, Obsidian, Mem, Capacities, Heptabase, Logseq, Roam, Evernote, etc. The official website and promotional copy often claim that "200,000+" creators, thinkers and professionals are using it; third-party app store data (such as mwm.ai) shows that the number of downloads is in the order of 50,000 and the score is about 4.9/5, but the App Store's ontology score of 5.0 is only based on 13 reviews. The sample is small and the reference value is limited.

  • Fabric strings "capture-organizing-retrieval-collaboration" into a chain. The capture end is very comprehensive: browser plug-in can cut web pages with one click, save directly in the mobile sharing panel, voice notes are transcribed by AI, screenshots are automatically stored in the database, files can be uploaded by dragging and dropping, and emails and news emails can be forwarded to a dedicated Fabric email address for instant archiving. The organization side focuses on "no organization": you don't need to create folders or tags. Fabric uses semantic understanding to classify content and automatically establish associations between data. When searching, you can describe the intention in vernacular without having to remember the file name. The search end has three levels - natural language semantic search (can understand the meaning of PDF, pictures, and audio transcriptions), visual/color search (describe "a photo of a jazz cafe with a retro atmosphere" to retrieve screenshots), and locate specific paragraphs across file types. The AI ​​assistant answers based on your own data. Each answer has a traceable reference source. It can modify documents, make summaries, and turn recordings into to-dos. Collaboration supports shared space, real-time co-editing, annotations and comments, and @mentions; it also provides canvas-style space view, tasks/reminders, and external publishing with analysis. External sources that can be connected include Gmail, Google Drive, Dropbox, Notion, Readwise, and more.

  • The current personal version on the official website is divided into four levels: Free (free, 150 item limit, 250MB, 25MB single file, AI limit), Plus (about 4.67 US dollars/month, annual payment of 56 US dollars, unlocks own AI agent, meeting records, intelligent memory, unlimited entries, 50GB), Pro (about 12.50 US dollars/month, annual payment of 150 US dollars, 2TB, heavy AI Usage, web search, automation, full model, link analysis, labeled "most popular"), and Teams. Older third-party information has appeared in Basic/Max and other earlier gears and different prices (such as Pro 20 US dollars, Max 50 US dollars), which are historical versions, and the current gears on the official website should prevail. The whole thing is a Freemium subscription system. The free version deliberately limits AI usage and connection capabilities. The core selling point "Connections/Associations" needs to be upgraded to experience it - this has been repeatedly criticized in user reviews.

  • The positive voices focus on several points: visual/color search is called by the creator "the first function that is truly like an extension of thinking"; voice notes are automatically transcribed, making it easy to take notes while commuting; the "self-organizing" experience allows people who have been hoarding information for a long time to finally "find it back"; some users praised the founder for being "passionate and constantly updating frequently." But the negative feedback is just as dense and specific. Multiple one-star reviews on the App Store pointed out that the iOS/iPad side is "frustratingly laggy": drafts will be lost due to weak network connections, there is no offline mode (completely unusable on a plane), the transcription will stop after 30 seconds, and the screen will jump to other notes inexplicably. Some users complained that the app automatically logged out frequently, causing clipping and sharing to fail. The highest paid tier was also accused of "unworthy of the name" in terms of AI usage, which bottomed out after just one or two weeks of use. Advertisements urging the purchase of more AI were everywhere; tags and smart collections were inaccurate, and customer service emails were silent. Chinese reviews (such as Guan Yu) are more pointed: the UI is empty, the AI ​​tags and dialogue intelligence are insufficient, the core function paywall blocks free users, and the overall "not worth paying".

  • In the "Second Brain" tool review, Fabric is often placed at the extreme of "zero maintenance": it advocates "just save, don't manage" and relies on the Memory Engine to automatically establish associations. Compared with Obsidian, Logseq, and Roam, which require manual construction of double chains and graphs, it is the most friendly to novices. However, industry observers also generally point out that the quality of its "automatic connection" is still unstable, and compared with Mem (text-based) and Heptabase (card-based whiteboard), it is not omnipotent in segmented scenarios. The technology media doesn’t report much on it, and more of the noise comes from personal blogs and the App Store. It belongs to the stage of “diehard fans in small circles and low public awareness.”

  • The most prominent risk is privacy and data ownership: Fabric needs to connect to a large number of personal data sources such as Gmail, Drive, Dropbox, Notion, etc., and declares transmission SSL and static AES-256 encryption, but "AI reading everything about you" itself is a threshold for privacy-conscious users; the App Store privacy label shows that it will collect contacts, identifiers, search history, user content, etc. and may be used for analysis and personalization. Secondly, there are reliability risks: a large number of one-star pointer synchronization failures, logouts, lost drafts, and AI limits. As the "second brain", once the link is lost at a critical moment, the trust cost is extremely high. The third is the risk of the business model: the free version has a low sense of value, a paywall blocks core functions, and complaints about AI usage design put it in the awkward position of "wanting to charge but unable to retain people." Finally, the product is still an early version of 0.8.x, and long-term survival and pricing strategies are subject to change.

  • Suitable for: creators, researchers, product managers and freelancers who have extremely scattered data, are tired of using note-taking systems, and want to "save and use"; especially suitable for people who rely heavily on visual/screenshot retrieval. Not suitable for: People who require offline availability, strong privacy local priority, or those who have already run the Obsidian/Logseq workflow smoothly. It is recommended to use the free file to connect commonly used sources and try it for a month, focusing on verifying whether semantic search and AI question and answer can really help you; don't rush to pay annually, wait until iOS stability improves and AI usage rules are more transparent before upgrading. Alternative solutions include Mem (organizing text to AI), Heptabase (whiteboard cards), Notion (all-purpose database), and the domestic Get Notes.

  • Fabric's vision is "a second brain that can think on its own." It does have unique advantages in terms of comprehensive capture, smart retrieval, and automatic correlation. In particular, its visual search and continuous updates are worthy of recognition. But it’s still too young: the iOS experience is awkward, offline lack, AI usage and paywall disputes have compromised the implementation of “Ideal is Full”. It's great to use it as an "online version of clipping + semantic search", but as an "irreplaceable second brain" you have to wait for a few more versions.

User Reviews

  • 头像
    bdrxwohj
    价格能接受,关键是别再砍功能。

  • 头像
    Stephanie_Roberts_Plus
    安卓端反而比 iOS 稳一些。

  • 头像
    猫咪_8
    同步偶尔抽风,希望稳住。

  • 头像
    BobbyLee_2023095
    创始人更新很勤快,这点加分。

  • 头像
    BlockKing_m
    当成第二大脑还太早,再等等。

  • 头像
    Jack_ParkerK
    最高档刚用一两周 AI 用量就见底,然后满屏都是催你加购 AI 的广告,收费设计让人很不舒服。

  • 头像
    TheGostyataSemenyak_88
    当成联网剪藏加语义搜索用很爽。

  • 头像
    z05pqwrdjj
    共享空间协作挺顺手的。

  • 头像
    heavyswan809
    自动登出这个毛病反复出现,一登出剪藏和分享就失效,关键时刻根本存不进去,信任感掉得一干二净。

  • 头像
    TheNoahStephens_88
    转写 30 秒就断,体验割裂。

  • 头像
    SatsChasera76
    视觉搜索是真香,创作者福音。

  • 头像
    游客918
    用了一个多月,最惊艳的是颜色搜索和截图检索,我描述「复古爵士咖啡馆」就能翻出半年前的截图,这点真的像思维延伸。

  • 头像
    Jesse.Butler_7
    链接分析功能对研究者很实用。

  • 头像
    gonbzen5z
    iOS 端体验真的拉胯,弱网会丢草稿,飞机上完全用不了,没有离线模式对「第二大脑」来说太致命了。

  • 头像
    CeylanKıraç
    客服邮件石沉大海,体验差。

  • 头像
    David_ThomasQ
    语音笔记转写太香了,通勤路上随便说几句,到工位已经变成带要点的文本,省了我大量整理时间。

  • 头像
    CLcoo
    界面有点空,空间利用率不高。

  • 头像
    ChristineClark_88182
    整体 idea 好,执行还差点意思。

  • 头像
    Grace_ThomasJr
    比 ChatGPT 记忆好用,能溯源。

  • 头像
    Zachary.Diaz_202364
    免费版几乎体验不到核心价值,核心的关联功能要升级,等于逼着先付费再判断值不值,门槛有点高。

  • 头像
    Angela_Phillips007
    Notion 同步只能同步主页,子页不行。

  • 头像
    HCampbell_Pro035
    数据都在云上,隐私党慎入。

  • 头像
    JArod
    中文评测说得没错,AI 自动打的标签准确度一般,智能集合经常对不上,整理这件事还是得自己兜底。

  • 头像
    EHall_Plus
    免费档先试试再说,别急着年付。

  • 头像
    lazyelephant147
    我之前 Notion、Obsidian、Roam 都搭过,最后都荒废了,Fabric 这种「只管存不用理」的思路才真正适合我这种懒人。

  • 头像
    HrithikBanerjee
    检索准得离谱,找回老资料毫不费力。