Give Claude Your Files and Watch What Happens

Give Claude Your Files and Watch What Happens ![rw-book-cover](https://i.ytimg.com/vi/HxEQ7bLKrqI/sddefault.jpg) 🔗 **Source:** [Give Claude Your Files and Watch What Happens](https://www.youtube.com/watch?v=HxEQ7bLKrqI) > [!note]+ Document Notes > tags: [ai-productivity, file-management, claude-cowork, automation, digital-organization] > author: [MISSING] > source: YouTube > reference: https://www.youtube.com/watch?v=HxEQ7bLKrqI > date created: {{date:YYYY-MM-DD}} {{time:HH:mm}} > > # 🏷️ Title: Give Claude Your Files and Watch What Happens >  ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ## 📝 Summary - **What happened**: Claude Co-work now offers direct file system access, allowing AI to read, edit, and organize files without manual upload workflows. - **Why it matters**: This fundamentally changes AI productivity workflows from chat-based assistance to workspace collaboration, enabling bulk document processing, cross-folder synthesis, and longitudinal analysis. - **Key stakeholders**: Content creators, researchers, knowledge workers, productivity system users (especially PARA Method practitioners). - **Risks and tradeoffs**: Privacy concerns with AI accessing entire file systems; potential over-reliance on AI for file organization; quality control challenges with bulk processing. - **Bottom line**: File system access transforms AI from a conversational tool into a workspace partner that can execute complex, multi-document tasks across entire digital archives. ## 📖 Main Content Claude Co-work's file system access represents a paradigm shift from upload-based AI interaction to shared workspace collaboration. The demonstration covers seven specific capabilities that weren't previously possible: **Parallel Multi-Document Processing**: Instead of uploading files one at a time, Claude can now process entire folders simultaneously. In the demo, it analyzes five PDFs in parallel to create a comprehensive research dossier, finding connections across papers and synthesizing themes. **Direct File Editing**: Claude can modify existing files rather than just providing suggestions. When asked to make content more beginner-friendly, it directly edits the markdown file in place, eliminating the copy-paste workflow. **New File Creation with Advanced Features**: The system can create sophisticated documents with features like tracked changes in Word documents. In the example, Claude reviews a 60,000-word manuscript and creates a revised Chapter 1 with detailed editorial suggestions as tracked changes. **File System Organization**: Claude can create folder structures, rename files and folders, and organize chaotic project directories. The AI course folder transformation shows it moving from a disorganized dump to a logical hierarchy with clear naming conventions. **Longitudinal Analysis**: Perhaps most impressively, Claude can analyze historical data across years of documents. The health folder analysis spans 18 years of medical records, identifying patterns and creating forward-looking recommendations that no single doctor consultation could match. **Existing Tool Integration**: The system works with tools like Obsidian by accessing local markdown files, surveying over 1,500 notes to assess whether there's enough material for a book project. **Cross-Folder Synthesis**: The most advanced capability involves working across multiple unrelated document collections simultaneously, such as combining project files with marketing area folders to create launch strategies informed by historical learnings. ### 🗒️ Notes #### ⏱️ [00:00:00](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=0) Introduction - File Upload Routine Elimination - Traditional workflow: find file, drag to chat, wait for upload - New paradigm: AI shares your file system and workspace - Seven specific real-world tasks demonstrated #### ⏱️ [00:01:15](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=75) Task 1 - Parallel Multi-Document Reading - Command: "Read and summarize all the PDFs in this folder" - Five PDFs processed simultaneously in 3-4 minutes - Output: 2-3,000 word research dossier with cross-paper synthesis - Includes connections to current work and highest leverage insights #### ⏱️ [00:03:45](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=225) Task 2 - Direct File Editing - Request to revise markdown document for Obsidian beginners - Completed in one minute with direct file modification - No new file creation - edits existing file in place - AI executes work rather than just providing instructions #### ⏱️ [00:05:20](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=320) Task 3 - New File Creation with Advanced Features - 60,000-word manuscript analysis focusing on Chapter 1 - 10-12 minute processing time for substantial revision - Creates new Word document with tracked changes - Comments from "Claude" reviewer with accessibility suggestions #### ⏱️ [00:08:10](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=490) Task 4 - File System Organization - Chaotic AI course folder with mixed file types - Request: organize into logical subfolders with clear naming - Creates hierarchical structure with renamed files for clarity - Flags potential duplicates for human review #### ⏱️ [00:11:30](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=690) Task 5 - Longitudinal Health Analysis - PARA Method structure: Projects, Areas, Resources, Archives - Health folder spanning over a decade of documents - 8-9 minute analysis identifying two distinct health chapters - Provides roadmap for sleep, energy, and longevity improvements #### ⏱️ [00:15:45](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=945) Task 6 - Obsidian Vault Survey - 1,500+ notes processed in 2-3 minutes - Book viability assessment for "competing through fun" concept - Identifies raw inputs vs. synthesized thinking gaps - Highlights relevant existing content and precedents #### ⏱️ [00:18:20](https://www.youtube.com/watch?v=HxEQ7bLKrqI?t=1100) Task 7 - Cross-Folder Synthesis - Combines AI course project with marketing and content area folders - 10-minute analysis for cohort 2 launch planning - Identifies dual buyer psychology and referral program insights - Draws on long-term learnings across multiple domains ## 📚 References - [Original video](https://www.youtube.com/watch?v=HxEQ7bLKrqI) - Building a Second Brain methodology (mentioned in context) - PARA Method (Projects, Areas, Resources, Archives) - Obsidian note-taking application ## 🗣️ Paraphrase Claude Co-work's file system access fundamentally transforms AI from a conversational assistant into a collaborative workspace partner. Rather than the traditional upload-and-chat workflow, AI now operates directly within your existing file structure, reading multiple documents simultaneously, editing files in place, and creating sophisticated outputs with advanced formatting features. The system demonstrates remarkable capabilities across seven key areas: parallel document processing that can synthesize insights across multiple sources; direct file editing that eliminates copy-paste workflows; creation of new documents with professional features like tracked changes; intelligent file organization that can restructure chaotic directories; longitudinal analysis that can identify patterns across years of historical data; integration with existing tools like Obsidian; and cross-folder synthesis that combines insights from unrelated document collections. This represents a shift from AI as a tool you use to AI as a collaborator that works within your existing digital environment, capable of executing complex multi-document tasks that would take humans hours or days to complete manually. ## 📰 Newsletter **AI Gets File System Access - Your Productivity Workflow Just Changed Forever** Remember dragging files into ChatGPT one by one? That's officially old school. Claude Co-work now works directly with your file system, and the implications for WordPress developers and content creators are massive. Imagine pointing AI at your entire plugins folder and asking it to audit security across all your custom code. Or having it analyze years of client project files to identify patterns in scope creep and budget overruns. One creator processed 1,500 Obsidian notes in minutes to assess if he had enough material for a book - a task that would have taken days manually. The real game-changer? Cross-folder synthesis. You could combine your WordPress development resources with your client communication archives to create better project templates, or merge your plugin documentation with user feedback to identify feature gaps. Early access is rolling out, but start preparing now by organizing your file structures. The PARA method (Projects, Areas, Resources, Archives) shown in the demo works brilliantly for this kind of AI collaboration. **Related**: Check out the [WordPress CLI documentation](https://developer.wordpress.org/cli/commands/) for file management commands that could complement AI-powered organization workflows. ## ✅ Action Items #### DevRel - **Developer advocacy**: Create demos showing file system AI integration with developer workflows - **Education**: Develop workshops on AI-assisted code review and documentation generation - **Developer experience**: Document best practices for AI-readable file organization - **Developer marketing**: Produce content showing competitive advantages of AI-integrated development #### WordPress - **Core**: Investigate AI integration opportunities for WordPress file management - **Documentation**: Audit existing docs structure for AI-readability optimization - **Training**: Create courses on AI-assisted WordPress development workflows - **Community**: Share best practices for AI-enhanced plugin and theme development #### Community - **Maintainers**: Develop AI-assisted code review and documentation workflows - **Contributors**: Create templates for AI-enhanced contribution processes - **Ecosystem**: Establish standards for AI-readable project organization ## 🧪 Hypotheses 1. **File system AI access will reduce content creation time by 60-80%** for knowledge workers managing large document collections 2. **Cross-folder synthesis capabilities will improve decision-making quality** by surfacing patterns across previously siloed information 3. **Direct file editing will eliminate 90% of copy-paste workflows** in AI-assisted content creation 4. **Longitudinal analysis capabilities will create competitive advantages** for professionals with extensive historical archives 5. **File organization automation will reduce time spent on information management** by 70-90% for chaotic digital workspaces ### 📊 Methodology - **Hypothesis 1**: Time-tracking studies comparing traditional vs. AI-assisted content workflows across document collections of varying sizes - **Hypothesis 2**: Decision quality assessments using expert panels to evaluate strategies created with and without cross-folder synthesis - **Hypothesis 3**: Workflow analysis measuring copy-paste actions and completion times in controlled editing tasks - **Hypothesis 4**: Longitudinal studies tracking business outcomes for professionals adopting historical data analysis - **Hypothesis 5**: Before/after productivity measurements in file management tasks with randomized control groups ### ❓ Questions - How will file system AI access change information security requirements for organizations? - What file organization standards will emerge to optimize AI collaboration? - How will this technology affect the competitive landscape for knowledge work? - What new privacy frameworks will be needed for AI file system integration? - How will traditional document management systems adapt to AI-native workflows? ### 🔮 Future Research - **Privacy-preserving AI file system architectures** that maintain functionality while protecting sensitive data - **Optimal file organization patterns** for AI collaboration across different professional domains - **Cross-platform AI file system standards** enabling seamless AI collaboration across different tools - **AI-assisted information governance frameworks** for enterprise file system management - **Long-term productivity impacts** of AI-native file management workflows ### 💡 Implications - **Practice**: Knowledge workers will need to restructure file organization for AI collaboration, potentially adopting methodologies like PARA - **Policy**: Organizations will require new governance frameworks for AI file system access and data privacy - **Security**: File-level AI access creates new attack vectors and requires updated security protocols - **Education**: Professional development will need to include AI collaboration skills and file organization best practices - **Technology**: Software tools will need to adapt to AI-native workflows, potentially changing fundamental UX patterns ### 🧠 Learning Styles Assessment **Reading level estimate**: Flesch-Kincaid Grade 10.2 - accessible to general audiences while maintaining technical depth #### 👀 Visual learners - **Present**: Screenshot demonstrations of before/after file organization, visual workflow comparisons - **Improve**: Architecture diagrams showing AI-file system integration, flowcharts of the seven task types #### 🔊 Aural learners - **Present**: Video format with narrated demonstrations - **Improve**: Audio-only podcast version, structured as step-by-step tutorials #### 🗣️ Verbal learners - **Present**: Clear definitions of technical terms, analogies comparing old vs. new workflows - **Improve**: Glossary of AI productivity terms, comparison charts between different AI file access methods #### 🧩 Kinesthetic learners - **Present**: Step-by-step task demonstrations with clear actions - **Improve**: Hands-on checklists for setting up file system AI access, practice exercises for each task type #### 🧠 Logical learners - **Present**: Structured seven-task framework, clear progression from simple to complex capabilities - **Improve**: Decision trees for when to use each task type, ROI calculations for AI productivity gains #### 🧑‍🤝‍🧑 Social learners - **Present**: Demonstration format suggests collaboration and sharing - **Improve**: Discussion prompts about team AI adoption, collaborative file organization strategies #### 🧘‍♀️ Solitary learners - **Present**: Self-directed video learning format, individual productivity focus - **Improve**: Self-assessment quizzes, personal file organization audit worksheets #### 🌿 Naturalistic learners - **Present**: Organic file organization that mimics natural hierarchies - **Improve**: Analogies to ecosystem organization, sustainable AI productivity practices ### 📊 Accessibility Assessment #### Screen reader semantics - **Present**: Video has clear narration describing visual elements - **Improve**: Detailed alt text for screenshots, structured headings for screen reader navigation #### Keyboard navigation - **Present**: Video player supports standard keyboard controls - **Improve**: Keyboard shortcuts for section navigation, transcript with jump links #### Contrast and readability - **Present**: Clear visual contrast in demonstration screenshots - **Improve**: High-contrast mode options, customizable text sizing in transcript #### Text resizing and reflow - **Present**: Video scales appropriately at different zoom levels - **Improve**: Responsive transcript layout, scalable interface elements #### Media and captions - **Present**: Video format with clear audio narration - **Improve**: Closed captions, downloadable transcript, audio descriptions of visual elements #### Forms and interactions - **Present**: Simple video player interface - **Improve**: Accessible video controls, clear focus indicators #### Standards reference - Maps to WCAG 2.2 AA criteria for multimedia content and navigation ### 🌍 DEIB Assessment #### Cultural breadth - **Present**: Productivity concepts applicable across cultures - **Improve**: Examples from diverse professional contexts, consideration of different work styles globally #### Inclusive language - **Present**: Professional, accessible language avoiding technical jargon - **Improve**: Plain-language alternatives for technical terms, multilingual transcript options #### Neurodiversity support - **Present**: Clear structure with numbered tasks, consistent demonstration format - **Improve**: Executive function support tools, attention management features in content delivery #### Representation - **Present**: Focus on productivity benefits accessible to various roles - **Improve**: Examples showing diverse professional backgrounds, inclusive imagery in supplementary materials #### Psychological safety - **Present**: Non-judgmental approach to current file organization methods - **Improve**: Acknowledgment of different comfort levels with AI technology, optional adoption pathways ### ✅ Logical Validity #### Strong reasoning - **Demonstration-based evidence**: Each task is shown working in real-time, providing empirical support for capabilities claims - **Progressive complexity**: The seven tasks build logically from simple (file reading) to complex (cross-folder synthesis) - **Practical application focus**: Each capability is grounded in real workflow problems rather than theoretical benefits #### Fallacies - **Cherry-picking**: The demonstration may show only successful cases without addressing failure modes or limitations - **Hasty generalization**: Extrapolating from a few demo tasks to broad claims about productivity transformation - **Appeal to novelty**: Presenting file system access as inherently better primarily because it's new technology #### Tone of voice - **Enthusiastic and promotional** with examples like "That's over now" and "just mind-blowing," suggesting marketing rather than neutral technical documentation ### 🛡️ DARVO Analysis - **Deny**: No evidence of denial tactics - the content straightforwardly presents new capabilities - **Attack**: No attacks on critics or alternative approaches - **Reverse victim/offender**: No victim/offender dynamics present in this technical demonstration This content appears to be straightforward product demonstration without DARVO elements. ## 🎯 Conclusion Claude Co-work's file system access represents a fundamental shift from AI as a conversational tool to AI as a workspace collaborator. The seven demonstrated capabilities - from parallel document processing to cross-folder synthesis - show potential for dramatic productivity improvements in knowledge work, particularly for professionals managing large document collections or complex project archives. **Practical next steps for readers**: 1. Evaluate your current file organization for AI-readiness 2. Consider adopting structured methodologies like PARA for better AI collaboration 3. Identify high-value use cases in your workflow that could benefit from bulk document processing 4. Prepare privacy and security frameworks for AI file system access 5. Start experimenting with file-based AI workflows as access becomes available The technology promises to eliminate much of the friction in AI-assisted productivity work, but success will depend heavily on thoughtful file organization and clear understanding of appropriate use cases. ## 🔄 Refactor Appendix — Give Claude Your Files and Watch What Happens ### 📚 Literature Atoms (H3) #### AI file system access eliminates upload workflows entirely > [!quote] ID:rw_demo1 > You find the file you want to work with. You click and drag into a chat window. You wait for it to upload. That's over now. **Context & use**: - Opening statement of video demonstration - Represents fundamental workflow paradigm shift from manual file handling **Tags**: #ai-productivity #workflow-automation #xref #### Cross-folder synthesis enables unprecedented strategic insights > [!quote] ID:rw_demo7 > You can actually compare and contrast and interrelate and interconnect or even merge and synthesize whole different collections of documents that may have little to do with each other **Context & use**: - Most advanced capability demonstrated (task 7) - Combines AI course project with marketing/content areas for launch planning **Tags**: #strategic-analysis #cross-domain-synthesis #xref #### Longitudinal file analysis surpasses traditional professional consultation > [!quote] ID:rw_health > This is a depth and breadth of health insight that was just not available before. None of my doctors has the time to dive this deep into that long-term arc **Context & use**: - 18-year health document analysis producing comprehensive roadmap - Demonstrates AI's advantage in processing historical archives **Tags**: #longitudinal-analysis #healthcare-ai #professional-augmentation #xref ### 🧠 Permanent Candidates (H4) #### AI workspace collaboration transforms knowledge work productivity fundamentally **Claim**: File system AI access shifts AI from conversational assistance to executable workspace collaboration, enabling bulk processing and historical analysis impossible with traditional workflows. **Evidence pointer**: [[#AI file system access eliminates upload workflows entirely]] **Scope & conditions**: Requires organized file structures; effectiveness depends on document quality and clear task specification **Links**: - [[Productivity automation]] — because extends beyond single-task automation - [[AI augmented decision-making]] — because applies-in strategic planning contexts  ### ✏️ Fleeting Queue - Historical document analysis patterns for identifying professional development trends — investigating personal archive mining strategies #fleeting #inbox - File organization standards for AI collaboration — researching PARA method adaptation for AI workflows #fleeting #inbox ### → Things 3 (max 2 actions) - [ ] Audit current project file structures for AI-readiness and reorganize using PARA methodology — productivity optimization #things - [ ] Research Claude Co-work early access availability and privacy implications — ai-tools evaluation #things --- ## Full Transcript [00:00] I know you've done this routine a [00:01] thousand times now. You find the file [00:04] you want to work with. You click and [00:05] drag into a chat window. You wait for it [00:07] to upload. [00:10] That's over now. AI can now read your [00:14] files. You and your AI share the same [00:17] file system, the same workspace. Let me [00:20] show you everything that changes. I'll [00:24] take you through seven specific [00:26] realworld tasks that I could not [00:28] accomplish without it having access to [00:31] my file system. And the first task that [00:34] I want to show you is that Claude or [00:37] really any AI that has this capability [00:40] can now not only read files but read [00:42] many files in parallel. So, I'll go [00:44] ahead and say, "Read and summarize all [00:47] the PDFs in this folder and give me a [00:50] research dossier of the main findings [00:53] and points and ideas that I could use in [00:56] my current projects." So, you can see [00:58] right there, there's five PDFs and it's [01:00] going to read them all in parallel. [01:05] All right, so that took around 3 or 4 [01:07] minutes and you can see I can click this [01:09] link right here to open up the document [01:11] it's created on the right. You can see [01:13] it's pretty extensive. It's maybe 2 or 3 [01:15] thousand words. It often kind of [01:17] summarizes the TLDDR over here on the [01:19] left in the chat and it's finding some [01:22] connections to my work with building a [01:24] second brain for my current book, Life [01:26] and Perspective that I'm writing and [01:28] then some notes on what it thinks are [01:30] the highest leverage frame. So, that's [01:32] helpful. So you can see here it's [01:33] created a brand new document in markdown [01:36] format uh summarizing everything that it [01:39] put in the research dossier and then [01:41] surfacing the connections to my current [01:43] work just as I asked. It's also doing [01:45] cross paper synthesis so different [01:47] themes that are found across the [01:49] different studies that I've collected. [01:51] And it even has a quick reference guide. [01:54] instead of having to work with one [01:56] document at a time, you can kind of go [01:58] up a level and work with entire [02:00] collections of documents. [02:03] So, what I want to show you now is that [02:05] co-work can directly edit an existing [02:09] file. So, I'm going to say for the [02:12] markdown document in this folder, go [02:15] ahead and revise it to be much more [02:18] beginner centric for people that don't [02:21] know anything about Obsidian. [02:23] And there we go. That took just about a [02:25] minute. You can see the main changes [02:26] that it made right here. And if we go [02:29] over here to that same folder we were [02:32] working in, you'll notice that it didn't [02:34] create a new one. It can actually go in [02:36] and change. It can edit and modify the [02:39] existing file. AI can now do the work. [02:42] It can execute the work rather than just [02:45] telling you in a chat what to do, but [02:47] then you are responsible for actually [02:49] making it happen. What I want to show [02:51] you now is that Co-work can not only [02:54] edit existing files, it can create brand [02:56] new ones. So, I'll go ahead and create a [02:58] new task here and choose a different [03:01] folder. Back to my desktop. And this is [03:03] a folder that contains the full draft, [03:06] the current working draft with the [03:08] manuscript of the book that I'm writing [03:10] called Life in Perspective. So, this is [03:12] a massive document. It's 60 or 70,000 [03:14] words. So, I'll go ahead and click open [03:16] and then always allow as usual. And then [03:18] I want to unselect the folder that we [03:21] were working in currently, Obsidian 101 [03:23] Workshop, so that we're only looking at [03:25] one. And then I'll just ask it, review [03:28] the current working manuscript for the [03:30] book I'm writing. Focus your attention [03:32] on chapter 1 and create a brand new [03:35] document containing chapter 1 with all [03:38] of your suggestions and recommendations [03:41] for improving it and making it more [03:43] accessible to an everyday reader. [03:46] include all your comments and [03:48] suggestions as tracked changes so that I [03:50] can review them individually. [03:57] That took around 10 or 12 minutes. So, [04:00] it was really a substantial task. I'm [04:02] going to go ahead and click show in [04:04] folder. And you can see it's created a [04:06] brand new document which I can open in [04:08] Word. And then we can see right here [04:10] that it has tracked changes with the uh [04:13] commenter Claude saying exactly what it [04:17] recommends I do to make this more clear [04:19] and more accessible. All right. The [04:21] fourth kind of task that's been unlocked [04:24] is that AI tools with file system access [04:28] can now creating new folders, moving [04:30] items into those folders, renaming the [04:33] items themselves, renaming the folders [04:35] themselves, it can create the kind of [04:38] structure, the kind of hierarchy that is [04:40] best suited for its needs and yours. Let [04:43] me show you what that looks like. So, [04:44] here I have a folder on my computer for [04:47] a recent course that I just taught uh [04:49] about AI. Actually, as happens with many [04:52] kind of large scale complex projects, [04:55] what we're left with is this kind of [04:56] like dump of just all sorts of things. [04:59] Logos, Google Docs, AI prompts, [05:03] spreadsheets, all sorts of different [05:05] things. So, what I'll do is start a new [05:08] task and go to choose a different [05:10] folder. Back to the desktop, choose the [05:13] AI course folder and open. Always allow. [05:18] Then I'll unselect the previous one that [05:20] we were working with and I'll just say [05:23] go ahead and organize this entire folder [05:28] into a series of subfolders that make [05:31] sense uh and are clear based on the [05:34] contents of the documents. Title the [05:37] subfolders clearly so that in the future [05:39] we'll know what's in them. also retitle [05:42] the documents themselves in case any of [05:45] them are unclear so that we can tell at [05:47] a glance what they contain. And there we [05:50] go. So it's summarized for me what the [05:53] final structure looks like. So here's [05:55] the top level folder and then the main [05:58] subfolders and then the sub subfolders [06:00] within those which makes sense. And then [06:03] these are all the files that have been [06:05] renamed for clarity. And then it's [06:07] flagging one thing to check, which is a [06:10] duplicate of the same file twice. So, [06:13] let's head over to the finder. And [06:16] voila, the exact same random pile of [06:20] stuff that we were looking at just about [06:22] 5 or 6 minutes ago is now a nice, neat, [06:25] organized set of documents. So, what I [06:29] want to show you here is basically that [06:30] AI is not just a personal assistant that [06:34] is executing specific low-level tasks [06:37] that you are giving it. It can actually [06:40] create workspaces. It can modify those. [06:42] It can reshape those as your work [06:44] evolves. The fifth kind of task that [06:46] gets unlocked is that you can iterate [06:49] over time. One thing you'll notice when [06:51] you work with AI is that it has no sense [06:53] of time, has no sense of history. [06:56] Everything from projects to areas of [06:58] your life to personal interests to how [07:01] your family is doing, everything has [07:03] this kind of evolution, this trajectory [07:06] that takes place over time. A lot of [07:08] that timeline, that trajectory is [07:11] reflected in something like a file [07:12] system. And often our file systems go [07:14] back many years. Let me show you an [07:17] example of what I mean. I'll go ahead [07:19] and create a new task. And what I'm [07:21] actually going to do now is instead of [07:23] giving it a project folder, which is [07:25] what we've been doing until now, I'm [07:27] going to go here in the sidebar. You can [07:30] see that I have my sidebar organized by [07:33] parah. I have my inbox. I have my [07:35] projects right here. I have my areas and [07:39] I have my resources and my archives. So [07:42] all of them are just one click away. And [07:44] this entire parah structure is contained [07:48] within my drive which means that it's [07:50] all synced to the cloud. It's synced to [07:52] Google Drive. So I have access to [07:54] everything I need both locally on my [07:56] computer where it's all mirrored as well [07:58] as on the cloud. So what I'm going to do [08:01] is click into areas. And if I click into [08:04] my health area, you can see I have all [08:06] sorts of things. I have health insurance [08:09] information, blood work labs, dental [08:12] content, life insurance, tests I've [08:14] done, juicing regimens, workout [08:17] routines, and these go back, you can't [08:20] tell by the dates right here, but these [08:22] go back many years, probably go back [08:24] over a decade. So, I'm going to go ahead [08:28] and click the health folder, give it [08:30] access to that. Then I'll unselect the [08:33] folder we were working in currently for [08:34] the AI course. And then I'm going to ask [08:37] it analyze the contents of this folder [08:40] which contains documents related to my [08:43] health over the years and give me a [08:46] picture of how my health has changed [08:50] over time. What are the main issues that [08:52] I have had, the challenges, the [08:55] improvements to my health that I've [08:57] realized? and then give me a road mapap [08:59] for areas that I could tackle going [09:01] forward so that I have better sleep, [09:04] more energy, uh, and more longevity for [09:07] the long term. [09:14] All right, so that took another good 8 [09:16] or 9 minutes and here's what it came up [09:18] with. So the documents that it had [09:21] access to cover 18 years of my health [09:24] history. So, it notes that my health [09:26] history over those 18 years has two [09:28] distinct chapters. The first was my ENT [09:32] voice crisis, which I've talked a lot [09:34] about publicly, which was indeed a [09:37] decadel long struggle with multiple [09:39] misdiagnosis, [09:41] a lot of cognitive side effects, real [09:43] psychological and social damage. It's [09:45] really an awful chapter of my life. And [09:47] I did, as it notes here, come out the [09:50] other side. So, this is more relevant. [09:53] The second chapter still unfolding is [09:55] about cardiovascular and metabolic [09:58] health. So it notes that my 2025 blood [10:01] work which is only about a year old [10:03] shows a significant deterioration in my [10:05] triglycerides, my HDL. I'm not going to [10:08] open up the full analysis and road map [10:10] because it contains some private [10:12] information. But at first glance just [10:15] based on this summary, this is just [10:16] incredible. This is a a both a depth and [10:20] also a breadth of health insight that [10:24] was just not available before. I mean, [10:26] I've had many doctors, many excellent [10:28] doctors, but none of them has the time [10:31] to dive this deep into that long-term [10:34] arc of how my health is changing. And I [10:36] think what I would do with this road map [10:38] is really dive in and start to identify [10:41] specific changes in my habits, in my [10:43] sleep, in my diet. Clearly, it's already [10:45] pointed that out. so that I can start to [10:47] address these things. [10:50] You can stay in your existing tools. So, [10:53] this is a huge deal. Let me show you [10:56] what I mean. I'll go ahead and open a [10:58] new task and then click to enter the [11:01] file system. And what I'm going to do [11:03] this time is go into my resources folder [11:06] and find my Obsidian vault. So, this [11:08] works with different kinds of apps, but [11:11] Obsidian is really one of the best [11:12] examples. And the reason for that is it [11:15] keeps its database of notes not in some [11:18] cloud or in some proprietary format but [11:21] in an open format which is markdown on [11:24] your local machine. So it's kind of AI [11:26] ready. So what I'm going to do is click [11:29] open unselect the previous folder. So [11:32] let me go ahead and ask it. Here is my [11:34] Obsidian Vault. do a general survey of [11:36] its contents to tell me whether I have [11:39] enough interesting and relevant material [11:42] for a book that I'm considering writing [11:45] which is expanding on the idea that you [11:49] can't compete with someone who's having [11:52] fun. So in other words, I want to [11:54] explore the role of pleasure and joy and [11:58] enjoyment and enthusiasm and fun not [12:02] just as ends in themselves though they [12:05] are but as a means to a higher level of [12:09] performance than is possible through [12:11] self-sacrifice and grinding. [12:20] So, that took only two or three minutes [12:23] and it looked at over 1,500 notes. That [12:26] alone again is just mind-blowing. How [12:29] long would it have taken me to go [12:31] through even just scanning 1,500 plus [12:35] notes? I mean, hours, days. Its short [12:38] answer is yes. There is enough material [12:40] and in some areas it's remarkably rich. [12:43] So, that's good to hear. Although I know [12:45] that AI has a strong tendency to be [12:47] psychopantic to tell me what I want to [12:49] hear. So I'm looking for evidence that [12:52] what it just said is true. So the vault [12:55] is stronger on raw inputs than on my own [12:58] synthesized thinking. That is very true. [13:00] I tend to mostly take notes just on the [13:02] raw inputs and not store the synthesized [13:06] thinking when I do it in the notes [13:08] themselves. So that does make sense. [13:10] highlights from a couple books on flow [13:12] states, peak performance, [13:13] neurochemistry, [13:14] positive emotion, [13:17] uh the broaden and build theory, which I [13:19] don't even remember having read about [13:21] with citations on the the benefits of [13:24] positive emotion. Notes from the book [13:26] satisfaction, which I do remember [13:28] reading. Interesting. So Ali Abdal's [13:30] book feelood productivity is the most [13:33] obvious published precedence for your [13:35] thesis. I hadn't even considered that, [13:37] but now I do see the connection. Okay, [13:39] so the bottom line is that what you'd be [13:42] doing next is the synthesis work, which [13:44] I totally agree with. I think that's [13:45] absolutely right. This is my favorite [13:47] one of all. The seventh and I would say [13:50] most sophisticated, most advanced kind [13:52] of task that is now available because of [13:56] file system access is crossfolder [13:59] synthesis. Because we can now go up a [14:01] level from the level of individual [14:03] documents to entire collections of [14:05] documents, we can also work between [14:09] different collections of documents. You [14:11] can actually compare and contrast and [14:14] interrelate and interconnect or even [14:17] merge and synthesize whole different [14:20] collections of documents that may have [14:23] little to do with each other or on the [14:25] surface have nothing to do with each [14:27] other. Let me show you a few examples. [14:30] So I'll go ahead and start a new task. [14:32] I'll say work in a project. Choose a [14:34] different folder. And I will first give [14:37] it the AI course folder. So has all the [14:40] contents of this specific project. But [14:43] then I also I can click again. I also [14:45] want to choose a different folder and [14:48] select within areas my marketing folder [14:52] which is an area as you can see here as [14:54] well as my content. So our content is [14:57] the main way that we do marketing. So I [14:59] want it to work across both content and [15:01] marketing and hit open. So you can see [15:04] here it now says AI course plus two. So [15:07] if I click what that means is that it's [15:10] drawing on the current project the AI [15:14] course plus the content area folder plus [15:16] the marketing area folder all three. So [15:19] now I can say analyze the contents of [15:21] the AI course folder which contains [15:24] everything related to our program the AI [15:27] second brain. Then create a launch and [15:30] marketing plan for cohort 2 which is [15:33] happening in fall 2026 and draw on the [15:37] contents of two area folders that I'm [15:40] also giving you. The marketing area [15:42] folder and the content area folder. [15:46] suggest any ideas, possibilities, [15:50] new strategies or tactics that we could [15:52] use to reach our audience and new [15:54] audiences based on the long-term [15:58] learnings from those two area folders. [16:07] All right, so that took a good 10 [16:08] minutes. It summarizes findings that [16:11] there's two different buyer psychology [16:13] modes. one person who has been on our [16:15] email list for years and another kind of [16:19] person that has been on our list less [16:20] than a month. That's super interesting [16:22] that we need to speak to both of those [16:24] two groups simultaneously. Uh the master [16:27] prompt template is doing double duty. So [16:28] we should uh double down on it as the [16:31] most resonant concept. The bring a [16:33] friend referral program had a 72% [16:36] redemption rate which is fantastic. So [16:37] we should really double down on that. So [16:40] this is super interesting. I can already [16:41] tell even by this very high-level [16:44] summary that it is drawing on just a a [16:48] bigger picture, a wider view of what [16:51] we're trying to accomplish through this [16:53] program beyond just this one single [16:55] launch that is happening over the course [16:57] of a few weeks. Today we covered file [17:00] system access which is just the first [17:02] thing that Claude Co-work offers above [17:05] and beyond chat. If you'd like a little [17:08] preview of what we'll be covering in [17:10] future parts of this series, check out [17:13] this video where I give you a broad [17:15] overview of what Claude Co-work is and [17:18] what it offers. And don't forget to hit [17:20] subscribe so you're sure to see the next [17:22] video in the series. And thanks for [17:24] watching.  📄 **Full text:** ![[Full Document Contents/Articles/Give Claude Your Files and Watch What Happens.md|Give Claude Your Files and Watch What Happens by YouTube]] ## Highlights - access to that. Then I'll unselect the folder we were working in currently for the AI course. And then I'm going to ask it analyze the contents of this folder which contains documents related to my health over the years and give me a picture of how my health has changed over time. What are the main issues that I have had, the challenges, the improvements to my health that I've realized? and then give me a road mapap for areas that I could tackle going ([View Highlight](https://read.readwise.io/read/01ktvd5gk661htw0mteqh5mhea)) - Tags: 👻 ai highlighted - So I'll go ahead and start a new task. I'll say work in a project. Choose a different folder. And I will first give it the AI course folder. So has all the contents of this specific project. But then I also I can click again. I also want to choose a different folder and select within areas my marketing folder which is an area as you can see here as well as my content. So our content is the main way that we do marketing. So I want it to work across both content and ([View Highlight](https://read.readwise.io/read/01ktvd5gm43phc0kfjb8nb5tjp)) - Tags: 👻 ai highlighted - marketing and hit open. So you can see here it now says AI course plus two. So if I click what that means is that it's drawing on the current project the AI course plus the content area folder plus the marketing area folder all three. So now I can say analyze the contents of the AI course folder which contains everything related to our program the AI second brain. Then create a launch and ([View Highlight](https://read.readwise.io/read/01ktvd5gndf7azp07stc55k69y)) - Tags: 👻 ai highlighted