12 Hidden Tricks to Level Up Claude Cowork

12 Hidden Tricks to Level Up Claude Cowork ![rw-book-cover](https://img.youtube.com/vi/76aUqqi-Gak/maxresdefault.jpg) 🔗 **Source:** [12 Hidden Tricks to Level Up Claude Cowork](https://www.youtube.com/watch?v=76aUqqi-Gak) > [!note]+ Document Notes > tags: #claude #ai-agents #productivity #workflow #automation > author: Frank > source: YouTube > reference: https://www.youtube.com/watch?v=76aUqqi-Gak > date created: 2024-12-28 15:22 > > # 🏷️ Title: 12 Hidden Tricks To Level Up Claude Cowork >  ## 🧰 Action Inbox - [H1] Migrate project instructions to claude.md files for portability — 01:45 #workflow - [H2] Set up folder-based organization with multiple folder access — 04:15 #organization - [H3] Enable voice dictation for unstructured brainstorming — 06:30 #productivity ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [Claude Cowork](https://claude.ai) - [Obsidian](https://obsidian.md) ## 📝 Summary - What happened: Frank shares 12 advanced techniques for maximizing Claude Cowork's capabilities beyond basic usage - Why it matters: Most users only scratch the surface of Claude Cowork's potential as an AI agent for non-coders - Key stakeholders: Claude Cowork users, AI productivity enthusiasts, non-technical professionals - Risks and tradeoffs: Over-complexity vs. efficiency, tool lock-in vs. portability, usage limits vs. functionality - Bottom line: Strategic setup and workflow optimization can dramatically improve Claude Cowork's effectiveness ## 📖 Main Content Frank presents a comprehensive guide to advanced Claude Cowork optimization, challenging conventional usage patterns. His core philosophy centers on future-proofing workflows by avoiding tool-specific lock-in while maximizing current capabilities. The tutorial emphasizes organizational strategies (using projects as folders rather than instruction containers), technical setup improvements (claude.md files for portability), and cost optimization techniques (off-peak scheduling, separate usage limits). Frank advocates for efficiency through skills and connectors while warning against feature bloat that can degrade performance. Key technical concepts include multi-folder access patterns, voice dictation workflows for unstructured input, and the integration of Obsidian as a file viewer for better output management. The approach balances immediate productivity gains with long-term workflow sustainability. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/76aUqqi-Gak?t=0) — [H1] Introduction to Claude Cowork optimization - Key idea: Most users only use a fraction of Claude Cowork's capabilities - Evidence: "Claude Co-work after receiving an incredible number of updates over the past few weeks is now one of the most powerful AI agents for non-coders" #### ⏱️ [01:00](https://youtu.be/76aUqqi-Gak?t=60) — [H1] Projects should organize chats, not store instructions - Key idea: Use projects as folder systems rather than instruction containers for future-proofing - Evidence: "Claude Co-work projects should only be used as a way to organize your chats and not as a way to give Claude instructions" #### ⏱️ [02:45](https://youtu.be/76aUqqi-Gak?t=165) — [H2] Claude.md files for portable instructions - Key idea: Store all instructions in claude.md files within working directories - Evidence: "This is just a text file that gives instructions to Claude co-work. Anytime it goes into this folder to do work, it will read this file" #### ⏱️ [04:15](https://youtu.be/76aUqqi-Gak?t=255) — [H2] Multi-folder selection for targeted context - Key idea: Select multiple folders to give Claude access to relevant information without overwhelming it - Evidence: "When you start a new task and choose which folder on your computer you want to work inside of, you can actually choose multiple folders" #### ⏱️ [05:00](https://youtu.be/76aUqqi-Gak?t=300) — [H3] Set default folders for efficiency - Key idea: Configure frequently used folders as defaults to streamline workflow - Evidence: "Set a default folder if you find yourself using one folder more than all the others" #### ⏱️ [05:15](https://youtu.be/76aUqqi-Gak?t=315) — [H2] Understanding dispatch limitations - Key idea: Mobile access through dispatch has significant practical limitations - Evidence: "Your computer needs to be on. Co-work needs to be open, and then you need to link up dispatch with it... it honestly doesn't work that well" #### ⏱️ [06:30](https://youtu.be/76aUqqi-Gak?t=390) — [H1] Voice dictation for brainstorming - Key idea: Use voice input for unstructured idea capture and let AI organize thoughts - Evidence: "A lot of times our best ideas come to us as we're talking out loud about them and just sort of rambling" #### ⏱️ [07:45](https://youtu.be/76aUqqi-Gak?t=465) — [H2] Schedule tasks during off-peak hours - Key idea: Avoid peak usage hours (5AM-11AM PST) to conserve usage limits - Evidence: "Peak hours, which are between 5:00 a.m. and 11:00 a.m. Pacific Standard Time, and that will consume your usage limits faster" #### ⏱️ [08:30](https://youtu.be/76aUqqi-Gak?t=510) — [H1] Prioritize skills over long instructions - Key idea: Use skills for efficiency rather than loading all instructions every time - Evidence: "Skills fix that because your agent will only load the skills that it needs or the ones that you tell it to load" #### ⏱️ [10:00](https://youtu.be/76aUqqi-Gak?t=600) — [H2] Leverage connectors for app integration - Key idea: Use connectors to create a central command center for multiple applications - Evidence: "If you find yourself jumping between lots of apps to get your work done, see if there's a connector for it" #### ⏱️ [10:30](https://youtu.be/76aUqqi-Gak?t=630) — [H2] Use Claude Design for creative tasks - Key idea: Utilize separate usage limits in Claude Design before consuming Cowork limits - Evidence: "Claude Design... is included with your Claude subscription... They're separate. So you might as well burn through those ones first" #### ⏱️ [11:15](https://youtu.be/76aUqqi-Gak?t=675) — [H2] Obsidian for file viewing - Key idea: Use Obsidian as a unified viewer for Claude Cowork outputs - Evidence: "Download and use the app Obsidian. It's free to use and this will be a viewer that you can use to view the files that Claude Co-work makes for you" #### ⏱️ [12:30](https://youtu.be/76aUqqi-Gak?t=750) — [H3] Regular cleanup of skills and connectors - Key idea: Monthly maintenance prevents performance degradation and confusion - Evidence: "If you have too many skills, especially ones that you're not even using, it can degrade the performance and push you closer to your limits faster" ## 📚 References - [Claude Cowork Platform](https://claude.ai) - [Obsidian Note-Taking App](https://obsidian.md) - [Frank's School Community Waitlist](URL in video description) ## 🗣️ Paraphrase - Claude Cowork projects work best as organizational folders rather than instruction repositories, ensuring workflow portability across AI tools - A claude.md file system provides portable, hierarchical instructions that can easily transfer to future AI platforms - Voice dictation combined with AI processing transforms unstructured brainstorming into organized, actionable outputs ## 📰 Newsletter Frank's comprehensive Claude Cowork optimization guide reveals advanced techniques that most users miss. His folder-based organization system and claude.md instruction files offer a future-proof approach that prevents tool lock-in while maximizing current capabilities. Key takeaway: Use projects as folders, not instruction containers. Store all guidance in portable claude.md files within your working directories. This setup transfers seamlessly to future AI tools while keeping your current workflow organized and efficient. The voice dictation feature particularly shines for brainstorming sessions - Claude excels at transforming rambling thoughts into structured output. Combined with strategic scheduling during off-peak hours and regular cleanup of unused skills, these optimizations can significantly extend your usage limits. Resources: [Claude Cowork](https://claude.ai), [Obsidian](https://obsidian.md), [AI Productivity Workflows](https://example.com) ## ✅ Action Items #### DevRel - Developer advocacy: Create tutorials showcasing claude.md file patterns and multi-tool workflows - Education: Workshop on AI agent optimization and workflow portability strategies - Developer experience: Documentation for file-based instruction systems - Developer marketing: Case studies of advanced Claude Cowork implementations - Community management: Forums for sharing optimization techniques and troubleshooting #### Community - Contributors: Templates and examples for claude.md instruction files - Sponsors: ROI analysis of AI agent optimization for enterprise workflows - Ecosystem: Integration patterns between Claude Cowork and other productivity tools ## 🧪 Hypotheses - File-based instruction systems improve workflow portability across AI platforms compared to tool-specific settings - Voice dictation input produces higher quality brainstorming outputs than text-only interaction - Strategic usage limit management through scheduling extends effective AI agent capacity - Skills-based architecture reduces token consumption compared to comprehensive instruction loading ### 📊 Methodology - A/B testing of project instruction methods vs. claude.md file approaches for workflow migration - Comparative analysis of voice vs. text input quality for unstructured ideation tasks - Usage analytics comparing peak vs. off-peak scheduling impact on limit consumption ### ❓ Questions - How do different instruction file architectures affect AI agent performance consistency? - What organizational patterns emerge in successful multi-folder Claude Cowork implementations? - How might voice dictation workflows evolve as AI audio processing capabilities improve? ### 🔮 Future Research - Cross-platform AI agent instruction portability standards and best practices - Voice-first AI interaction design patterns for knowledge work - Optimization frameworks for multi-AI-tool workflow orchestration ### 💡 Implications - Practice: Fundamental shift from tool-specific to portable AI workflow architectures - Policy: Need for standardized AI agent instruction file formats across platforms - Security: File-based instruction systems require careful access control and version management - Education: Training programs on AI workflow portability and optimization techniques - Funding: Investment in tool-agnostic AI productivity infrastructure ### 🧠 Learning Styles Assessment Reading level estimate: 9th-10th grade (technical tutorial format) #### 👀 Visual learners - Present: Screen recordings, folder structure diagrams, UI demonstrations - Improve: Workflow architecture diagrams, decision trees for optimization choices #### 🔊 Aural learners - Present: Video narration with clear explanations of each technique - Improve: Audio-only version with detailed verbal descriptions of visual elements #### 🗣️ Verbal learners - Present: Step-by-step explanations, analogies (folders as organization), technical definitions - Improve: Glossary of AI agent terminology, comparison charts between approaches #### 🧩 Kinesthetic learners - Present: Hands-on demonstrations, setup walkthroughs, interactive examples - Improve: Practice exercises, configuration checklists, troubleshooting guides #### 🧠 Logical learners - Present: Structured 12-point framework, cause-and-effect relationships - Improve: Decision matrices for optimization choices, performance metrics #### 🧑‍🤝‍🧑 Social learners - Present: Community references, shared workflows, collaborative optimization - Improve: Discussion prompts, peer review templates, group optimization sessions #### 🧘‍♀️ Solitary learners - Present: Individual optimization techniques, personal workflow customization - Improve: Self-assessment checklists, reflection questions on workflow efficiency #### 🌿 Naturalistic learners - Present: Organic workflow evolution, ecosystem thinking about tool integration - Improve: Sustainability considerations for AI usage, workflow lifecycle thinking ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear video narration describing on-screen actions - Improve: Detailed alt text for UI elements, structured headings in any written materials #### Keyboard navigation - Present: Demonstrations include keyboard shortcuts (Command+D for dictation) - Improve: Complete keyboard-only workflow alternatives #### Contrast and readability - Present: Clear screen recordings with good contrast - Improve: High-contrast mode demonstrations, text size considerations #### Text resizing and reflow - Present: Video format adapts to different screen sizes - Improve: Transcript with responsive formatting #### Media and captions - Present: Video with clear audio narration - Improve: Captions for all video content, downloadable transcript #### Forms and interactions - Present: Clear demonstration of UI interactions - Improve: Alternative interaction methods for accessibility tools ### 🌍 DEIB Assessment #### Cultural breadth - Present: Focus on universal productivity challenges - Improve: Examples from diverse work contexts, international time zone considerations #### Inclusive language - Present: Gender-neutral terms, accessible technical explanations - Improve: Plain-language alternatives for technical jargon #### Neurodiversity support - Present: Structured presentation, clear section breaks - Improve: Multiple ways to process information, accommodation for different attention styles #### Representation - Present: Focus on workflow efficiency over personal characteristics - Improve: Diverse use cases, varied professional contexts #### Psychological safety - Present: Encouraging tone about learning and optimization - Improve: Acknowledgment of different skill levels, learning pace variations ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Clear if-then relationships between optimization choices and outcomes - Inductive strength: Generalizations from specific techniques to broader workflow principles - Causal reasoning: Well-established connections between setup choices and efficiency gains - Analogical strength: Effective folder/organization metaphors for project structure #### Fallacies - Appeal to novelty: Positioning newer techniques as inherently better without full justification - False dilemma: Some either/or presentations could acknowledge middle-ground approaches #### Tone of voice - Overall tone: Confident and instructional with practical focus, occasionally promotional for community ### 🛡️ DARVO Analysis - Deny: No denial patterns present - Attack: No attacks on users or competing approaches - Reverse victim/offender: No victim/offender dynamics present - Emotional cues and rhetoric: Positive, empowering language focused on capability enhancement ## 🎯 Conclusion Frank's 12 optimization techniques transform Claude Cowork from a basic AI assistant into a sophisticated workflow orchestration platform. The key insight centers on future-proofing through portable, file-based instruction systems rather than tool-specific configurations. Practical next steps include implementing the claude.md file structure, experimenting with voice dictation for brainstorming, and establishing regular maintenance routines for skills and connectors. These optimizations compound over time, creating increasingly efficient AI-human collaboration patterns. The broader implication points toward an emerging need for standardized, portable AI workflow architectures as the ecosystem matures beyond single-tool solutions. ## 🧱 Refactor Appendix ### 📚 Literature Split Pad ### [H1] File-based AI instructions improve workflow portability > [!quote] [12 Hidden Tricks To Level Up Claude Cowork](https://www.youtube.com/watch?v=76aUqqi-Gak) by [[Frank]] via [[YouTube]]. Published 2024. Accessed 2024-12-28. > "To futureproof your setup, you want to make it as transportable as possible." Context: Advocating for claude.md files over project-specific instructions for easier migration between AI platforms. [view highlight](https://youtu.be/76aUqqi-Gak?t=165#H1) #xref #articleseed ID:rw_cowork_portability ### [H --- ## Full Transcript [00:00] Cloud Co-work after receiving an [00:02] incredible number of updates over the [00:03] past few weeks is now one of the most [00:05] powerful AI agents for non-coders. But [00:08] most people are only using a fraction of [00:10] its capabilities. If you just learn the [00:13] 12 simple Claude Co-work tricks that I'm [00:15] about to share with you though, you'll [00:17] finally unlock its full potential. And I [00:20] want to start with what is probably my [00:22] most controversial take. My first tip is [00:24] that Claude Co-work projects should only [00:27] be used as a way to organize your chats [00:29] and not as a way to give Claude [00:31] instructions. So, inside of Claude [00:33] Co-work, there are these things called [00:35] projects. Here are my projects. And if [00:38] you open up any of these, you'll notice [00:39] that over here on the right, you can [00:41] give the project instructions. The idea [00:44] behind this is that you can put some [00:45] instructions in here. So, every time you [00:47] start a chat from within a project, it [00:50] will automatically apply those [00:51] instructions. The problem is, as great [00:54] as Claude Coowwork is, we shouldn't fool [00:57] ourselves into believing that it will be [00:58] the best tool forever. To futureproof [01:01] your setup, you want to make it as [01:03] transportable as possible. But if you [01:06] have instructions inside of 20 different [01:09] projects, it's going to make it very [01:11] hard to move from co-work to another [01:14] tool. So, what do I use projects for? [01:17] Well, I only use them for one thing, [01:19] which is to organize chats. So, inside [01:22] of Cloud Co-work, every time you click [01:24] on new task, which is Co-work's name for [01:26] a new chat, and you start a task from [01:29] here, it will appear in the recent task [01:32] section here on the bottom left. But [01:34] this isn't organized, so it makes it [01:37] difficult to find important tasks that [01:39] you want to locate later. So, I've set [01:41] up projects that I use as folders. For [01:44] example, here is one I've called [01:46] presentations. Anytime I want to create [01:49] a new presentation, instead of clicking [01:51] on new tasks at the top, I instead click [01:54] on the presentations project, I tell it [01:58] what I want it to create for me. And [02:00] then that task will appear down here [02:03] inside the project. And any outputs it [02:06] creates, like the slide deck, will [02:08] appear in this output section. That way, [02:11] it's all neatly organized in one place. [02:14] So projects for me are basically a [02:16] folder system for tasks. But if I don't [02:19] use instructions in projects, how do I [02:23] tell Claude Co-work how I want it to [02:25] work? For that, I suggest you utilize [02:27] trick number two. As you already know, [02:30] Claude Co-works directly with files and [02:32] folders on your computer. Opening up my [02:35] Finder, this is the main folder that I [02:38] have Claude Co-work inside of. I've [02:41] called it Frankbbot. Inside of this [02:43] folder is a file called claude.md. [02:47] This is just a text file that gives [02:49] instructions to Claude co-work. Anytime [02:53] it goes into this folder to do work, it [02:55] will read this file and it will give it [02:58] all the context it needs to help me and [03:00] the general instructions for working in [03:02] this folder. Another way of putting it [03:04] is instead of putting all of your [03:07] instructions inside of the project [03:10] instructions, you take everything you [03:12] would put in there and you just put it [03:14] in a text file called claw.md. [03:18] That's it. And then within each of these [03:20] subfolders, which I basically use as [03:23] different projects, there's another [03:25] claw.md file with more specific [03:28] instructions for that subfolder. With [03:31] this setup, anytime you switch to a new [03:33] tool like Codeex, you can just tell it [03:35] to find all the claw.md files and to [03:37] change it to the name that it uses for [03:39] its instruction files and then it'll be [03:41] ready to go. I've made videos about this [03:43] system before and have received a ton of [03:46] great questions about it. So, I'm [03:48] actually planning to build a school [03:49] community where I'll share with members [03:51] a step-by-step guide on how to set this [03:54] up along with a lot of other AI [03:56] tutorials that I don't share here on [03:58] YouTube, live Q&As's, and a members [04:01] forum. If you want to find out more [04:02] about that or get on the wait list, all [04:04] the details are in the description down [04:06] below. Tip number three. Did you know [04:08] that when you start a new task and [04:10] choose which folder on your computer you [04:12] want to work inside of, you can actually [04:14] choose multiple folders? All you have to [04:16] do is select a new folder and you'll see [04:19] which ones are selected with this check [04:21] mark on the right. I find this really [04:23] helpful when I want the agent to have [04:25] access to information in two different [04:27] folders, but I don't want it to have [04:28] access to all the folders because then [04:31] it will have too much information that [04:33] may not be relevant to the task. So, for [04:36] instance, if I have it do some research [04:37] for me, I might have it pull that [04:39] research into the research folder, but [04:41] then I might ask it to create a [04:43] presentation for me about what that [04:45] research was. So, then I would give it [04:47] access to both the research folder and [04:49] the presentation folder. And while I [04:51] have this drop down open, tip number [04:53] four is to set a default folder if you [04:56] find yourself using one folder more than [04:58] all the others. So, I've set my default [05:01] as the parent folder that contains all [05:04] the other folders that I have my agent [05:06] work on. That one is called Frankbbot. [05:08] So, if I start a new task, it will [05:11] always select that one by default, but [05:13] if I want to, I can always change it. [05:15] Tip number five may also be a bit of a [05:17] controversial one, but you need to [05:19] understand the limits of dispatch. [05:22] Dispatch is a feature that's supposed to [05:24] give you mobile access to clawed [05:26] co-work, but there are two issues with [05:29] it. One, in order for it to work, your [05:32] computer needs to be on. Co-work needs [05:34] to be open, and then you need to link up [05:36] dispatch with it, and then it will work, [05:39] which isn't possible for everyone. They [05:41] can't just leave their computer on all [05:42] the time. And the second problem is that [05:45] it honestly doesn't work that well. It's [05:47] just like one continuous conversation [05:49] that you have with co-work. So, you [05:51] can't access all of your chats. It's [05:54] hard to have multiple projects going at [05:56] the same time. Like, for simple tasks, [05:59] it works fine, but whenever I try to use [06:01] it for multiple things, it just ends up [06:04] frustrating me and getting confused. Tip [06:06] number six I have for you is to use [06:08] voice dictation the right way. Inside of [06:12] Co-work, instead of typing in here, you [06:15] can use a voice dictation button. It's [06:17] right here. But you can also access it [06:20] just by pressing command D and it will [06:23] start listening to what you're saying. [06:25] But there's a secret hidden menu here on [06:27] this dropdown. You can select whether [06:29] you want to hold to record which is like [06:31] push to talk or you can toggle it. So if [06:34] I turn this off and now I press command [06:37] D, it is now toggled on and I don't have [06:39] to hold it and can just talk out loud. [06:42] And then once I'm finished talking then [06:44] press command D again and it will toggle [06:46] off dictation. Dictation is a very [06:49] important tool when it comes to using AI [06:51] agents like Claude Co-work because a lot [06:54] of times our best ideas come to us as [06:57] we're talking out loud about them and [06:59] just sort of rambling. And the AI agents [07:02] are really good at taking our [07:04] unstructured brain dumps and turning [07:07] them into a structured output. So, don't [07:09] be afraid to turn on dictation and just [07:12] brain dump a bunch of ideas and let Claw [07:15] do the work of sorting through all of [07:17] it. Tip number seven is a quick one, but [07:20] it has saved me a lot of money. When you [07:23] schedule tasks, and you should be using [07:26] these because this is a great way to [07:27] automate your work, you want to make [07:29] sure you schedule them for off- peak [07:31] hours. If you schedule them for peak [07:34] hours, you'll see a warning like this. [07:36] and opening it up, you'll see that it [07:38] tells you that you've scheduled this for [07:40] peak hours, which are between 5:00 a.m. [07:42] and 11:00 a.m. Pacific Standard Time, [07:44] and that will consume your usage limits [07:47] faster. So, make sure if it's possible [07:51] to schedule these for off- peak hours. [07:54] Another tip along the same lines is if [07:57] you have a task that you need to run, [08:00] but you know you don't have enough [08:01] limits for it, just set it up as a [08:04] schedule task to run later in the day [08:07] when you know your limits are going to [08:09] reset. Even if you're going to be away [08:11] from your computer, then it will just [08:12] automatically start and when you get [08:13] back to your computer, it will be done. [08:15] Tip number eight is to use lots and lots [08:18] of skills. You can think of skills as [08:21] prompts that you load into Claude [08:23] Co-work once so you never have to type [08:25] them out every time you want to use [08:27] them. Instead of using skills though, a [08:29] lot of people will create a very long [08:31] set of instructions either in their [08:33] claw.md file or worse inside their [08:37] project instructions. It may say [08:39] something like, "Every time I tell you [08:41] to triage my email inbox, I want you to [08:43] do XYZ. And then every time I tell you [08:45] to update my CRM, I want you to do ABC. [08:49] The problem with that approach is that [08:51] every time you use the agent, it will [08:52] read through all of those instructions, [08:54] which is very inefficient. Skills fix [08:57] that because your agent will only load [09:00] the skills that it needs or the ones [09:03] that you tell it to load. So, it's [09:05] incredibly efficient. And because of [09:07] that, you should be using less [09:08] instructions and more skills. For [09:12] example, if we return back to the [09:14] folders that I've set up for my agent, I [09:16] have a subfolder project folder here [09:18] called presentations. And inside of [09:20] that, I have a claw.md file, which again [09:22] are the instructions for the agent. But [09:25] if you look at this, literally all it [09:27] says is this project folder is used when [09:30] I want you to create onbrand slide decks [09:32] for me. And when I tell you to do that, [09:34] use the slide deck skill. That's [09:36] literally all it says to do. And then [09:39] inside of Claude, we have the slide deck [09:41] skill in here. So, it's only going to [09:43] load that when it needs to use it, which [09:46] is a lot more efficient. And tip number [09:48] nine is to use lots and lots of [09:51] connectors. If you find yourself jumping [09:53] between lots of apps to get your work [09:54] done, see if there's a connector for it. [09:57] What I'm trying to build for my school [09:59] community is a system where people can [10:01] log into one tool like claude co-work [10:03] and be able to control everything from [10:06] there like one central command without [10:08] jumping between all these different [10:10] apps. Tip number 10 is another one that [10:12] will hopefully also save you some money. [10:15] Take advantage of Claude Design. This is [10:17] included with your Claude subscription [10:20] that you use to access Claude Co-work. [10:22] And with this you can create slide [10:24] decks, you can create websites, you can [10:26] create motion graphics. So instead of [10:28] setting those up inside of Claude [10:30] Co-work, which is going to use those [10:32] usage limits, instead use the usage [10:35] limits that are for Claude design. [10:37] They're separate. So you might as well [10:39] burn through those ones first before you [10:42] start burning through the ones inside of [10:44] Claude Co-work, which you may want to [10:45] use for other tasks. Tip number 11 I [10:48] have for you is to download and use the [10:50] app Obsidian. It's free to use and this [10:53] will be a viewer that you can use to [10:56] view the files that Claude Co-work makes [10:59] for you. Like I was saying before, [11:01] everything it makes for you lives inside [11:03] of folders on your computer. So, this [11:06] isn't the most userfriendly way to view [11:09] what it's made. You can open up each [11:11] folder and then you have to use a [11:14] different app to actually view the [11:16] different outputs whether that's slide [11:19] decks or websites that's made for you or [11:21] even just text documents. So if you [11:24] download this app Obsidian, it's free. [11:26] It's available for iOS, Android, [11:27] Windows, Mac, Linux, then you get an [11:30] experience like this. This is the same [11:33] folder, my Frankbot folder. You can see [11:35] all the subfolders in here and it's a [11:38] lot easier to view what was created. So, [11:41] here's a text file and I can come in [11:43] here and add anything to it that I want. [11:46] And even if it makes something for you [11:48] that's HTML, like this slide deck it [11:50] made for me, you can view that in [11:52] Obsidian. Obsidian can get very [11:54] complicated very quickly. You don't need [11:57] to know 95% of what it can do. You [12:00] literally just download it, point it to [12:01] your folder, and use it like this. and [12:03] don't worry about all the extra [12:05] complicated stuff. And tip number 12 I [12:08] have for you might seem like a [12:09] contradiction, but it's really not. So, [12:11] I'd say about once a month, you want to [12:13] come into customize, look at all the [12:16] skills that you have turned on, and just [12:18] turn off the ones that you no longer [12:20] use. If you have too many skills, [12:23] especially ones that you're not even [12:25] using, it can degrade the performance [12:28] and push you closer to your limits [12:30] faster, your token limits. Same thing [12:33] with connectors. If you have too many [12:35] connectors that you're not even using, [12:38] then that can confuse Claude Co-work and [12:40] it may try to load a connector that you [12:42] don't even want it to use. So, just [12:44] disconnect anything you're not using or [12:46] turn off any skills that you're no [12:48] longer using. So, there you go. 12 [12:50] tricks for leveling up Claude Co-work. [12:53] Hopefully, you learned something new [12:54] today. If you did, let me know that by [12:56] giving this video a thumbs up. Then, if [12:58] you want to get on the wait list for my [12:59] new school community, make sure to do [13:02] that at the link in the description down [13:04] below. Otherwise, thanks so much for [13:06] watching the video and I'll see you in [13:07] the next one. Bye for now.  📄 **Full text:** ![[Full Document Contents/Articles/12 Hidden Tricks to Level Up Claude Cowork.md|12 Hidden Tricks to Level Up Claude Cowork by YouTube]] ## Highlights - your setup, you want to make it as transportable as possible. But if you have instructions inside of 20 different projects, it's going to make it very hard to move from co-work to another tool. So, what do I use projects for? Well, I only use them for one thing, which is to organize chats. So, inside of Cloud Co-work, every time you click on new task, which is Co-work's name for a new chat, and you start a task from here, it will appear in the recent task ([View Highlight](https://read.readwise.io/read/01krzztkhf6ykwbq8kzmcts5na)) - Tags: 👻 ai highlighted - section here on the bottom left. But this isn't organized, so it makes it difficult to find important tasks that you want to locate later. So, I've set up projects that I use as folders. For example, here is one I've called presentations. Anytime I want to create a new presentation, instead of clicking on new tasks at the top, I instead click on the presentations project, I tell it what I want it to create for me. And ([View Highlight](https://read.readwise.io/read/01krzztkhqtehvmj0ss10tx834)) - Tags: 👻 ai highlighted - then that task will appear down here inside the project. And any outputs it creates, like the slide deck, will appear in this output section. That way, it's all neatly organized in one place. So projects for me are basically a folder system for tasks. But if I don't use instructions in projects, how do I tell Claude Co-work how I want it to work? For that, I suggest you utilize trick number two. As you already know, Claude Co-works directly with files and ([View Highlight](https://read.readwise.io/read/01krzztkjaac3hwdf72bzvwdxn)) - Tags: 👻 ai highlighted