# 160 EASY Claude Cowork Setup (full beginner tutorial)

Show Notes

In this solo episode, I walk through the exact process of setting up Claude Cowork from scratch — the right way — using the free Cowork Onboarding plugin I built and made open source. I go step by step through the full onboarding flow, from picking your workspace folder and connecting tools, to building three core context files (About Me, Brand Voice, and Working Style), generating global instructions that tie everything together, setting up scheduled tasks like a daily morning briefing and quarterly context review, and running an optional security review to lock in sensible rules.

Timestamps

00:00 – Why Cowork is useless out of the box

02:22 – Starting the onboarding flow from scratch

03:30 – Choosing workspace folder and onboarding path

04:37 – Workspace structure and file access permissions

05:45 – Connecting Google Workspace tools (Gmail, Drive, Calendar)

06:56 – Building context files: About Me, Brand Voice, Working Style

10:58 – Linking existing brand materials from Google Drive

12:44 – Generating global instructions from your context files

15:00 – Skills and plugin recommendations from the marketplace

17:24 – Setting up scheduled tasks: morning briefing and quarterly review

19:46 – Running the optional security review

23:23 – Wrapping up: what we built and the two manual steps left

Key Points

• Most people skip the onboarding and then wonder why Cowork feels generic. The entire point of setting it up properly — building context files and global instructions — is so Claude never has to ask who you are again. Every session loads with that context automatically.
• The three context files (About Me, Brand Voice, Working Style) are the foundation of the whole setup. Without them, Cowork is writing responses for a stranger. With them, it knows your business, how you communicate, and how you like your deliverables structured.
• Global instructions are the piece most people miss. Once you paste your generated global instructions into Cowork's settings, every single session loads that context from the start — no repeating yourself, no re-explaining your preferences.

Links Mentioned

Free Cowork Onboarding plugin — download it to run the full guided setup described in this episode: https://return-my-time.kit.com/f00d78554c

Build with AI community — get the full onboarding plugin including the self-assessment, workflow audit, and custom skill blueprints tailored to your business: https://www.skool.com/buildwithai/about

WisprFlow — the talk-to-text tool used during the onboarding to answer Claude's questions faster: https://wisprflow.ai

Find Me on Social

X/Twitter: https://x.com/coreyganim

Instagram: https://www.instagram.com/coreyganim/

LinkedIn: https://www.linkedin.com/in/coreyganim/

YouTube: https://www.youtube.com/@coreyganim

What is Build With AI?

Most AI podcasts talk about what's possible. Build With AI shows you how it's done, live. Each episode, host Corey Ganim brings on entrepreneurs and operators who share their screen and build real AI automations, workflows, and tool setups right in front of you. No boring slides. Nothing that hasn't been battle-tested. You'll watch actual implementations get built from scratch so you can follow along and do the same in your business. If you're a non-technical entrepreneur who wants to put AI to work without becoming a developer, hit play and build along with us.

Corey Ganim: Okay, so the problem with Claude Cowork is that out of the box, it is a very dumb new employee. That's essentially how it acts. So in order to get it to a point where it understands you and your business, you've got to create context files. You've got to connect it to your tools and you've got to set it up properly the same way you would onboard a new employee. So here's everything that we're going to create today. First and foremost, we are going to give it a workspace folder structure. We're going to connect it to a real folder on our computer. Next, we're going to connect it to the tools that we're already using, whether it's Google Workspace, Notion, Asana, et cetera. Then we're going to build three very specific context files. We're going to build an about me document so that Claude knows everything about me. Next, we're going to build a brand voice profile, which tells Claude how we talk. And then lastly, we're going to build a working style document, how I approach tasks, the way that I like my documents to look, et cetera. After that, we're going to generate global instructions, which essentially tells Cloud how to treat all of the context files as a whole. Then we're going to go through skills and plugins, schedule tasks, and then we're going to have it conduct ⁓ an optional security review to review everything that we put together and make sure that it looks good. Now, all of this that we're doing here is made possible because of a plugin that I designed called a Cowork Onboarding. So out of the box, if you didn't have this plugin, you would have to literally go through step by step and know exactly what to ask Claude so that it does all of this for you. Luckily, we've made this plugin free and open source for you guys. So if you want to download this Cowork Onboarding plugin, you can go at the link in the description or the show notes and download that completely for free. The only thing that the Cowork Onboarding plugin that I'm giving you guys for free can't do is ⁓ essentially run you through a self-assessment, right? So that version of the plugin is only available to our community members. So what the self-assessment does is after you go through the full onboarding, it will interview you to essentially audit your existing workflows to produce an impact versus effort matrix, recommend exactly which skills you should build based on your specific workflows, and even build you the skill blueprints for those custom skills. And those are only available to community members. Let's go through and run this onboarding flow from scratch. So I'm already using Cowork. So it knows a little bit about me, but I'm going to go in and tell it, essentially run this from scratch as if I have no tools, no connectors, nothing installed. And this is a brand new instance of Cowork. So let's do that. And I'll come back once it's ready to go. All right. So now that I've told it, we want to start from scratch. It's basically saying, If you're on a pro plan, you want to save cowork for your most important work since it eats a lot more usage than regular cloud chat. And it's going to ask which workspace folder do we want to use for this session? So I'm going to pick a folder now. And then next, and this is important, it's going to ask which onboarding path I want to run. So if you use our onboarding plugin, you can either opt for the full setup, which runs you through everything I discussed at the beginning of this video or... You can opt for a quick start setup, which is only going to create a portion of those documents that we mentioned at the beginning. Now for the sake of this video, I'm going to go through the full setup. It's not actually going to take 45 minutes. Usually it's a lot quicker, but if you go through the full setup yourself, set aside 45 minutes just in case. Now it's going to ask which folder would I like it to use. So I'm going to choose the test folder that I created for this video and I'll be right back. So I selected a demo folder that I created for the sake of this video. And now the important next step is it's going to take a peek at the folder to confirm that it's empty. And what it's asking me to do is it's actually asking for permission to search through that folder. So that's one thing that's great about Cloud Cowork is it does have guardrails built in. It can't just go rogue, start poking through your documents, poking through your files. You're going to have to give it permission to do the things that it needs to do. So. Next, it's going to ask how structured do you want your workspace, full structure, minimal or custom? Let's do full structure, which is what it recommends. How should Claude handle your workspace folders? And in this case, it's going to ask me if I want it to have read only access to my source folders, read or write with confirmation, and then full read or write, which is not recommended for beginners. So I'm just going to go with pretty much all of the recommended options during this onboarding session, which is read only source folders. which is what it's recommending. So now what Cloud Cowork is saying is it's going to set up task tracking for the full onboarding so that we can track where we are in the process. Now, one thing you'll notice about Cloud Cowork, I'm actually going to move myself over here to the left so that you can see it, is in the top right we have essentially a progress bar. It's going over every task that we're going to accomplish during this workflow, one through 10. starting at building workspace, next connect tools, context files, all the way to step 10, which is verifying the onboarding completeness. Now this is a cool part about cowork because you can see exactly where it is in the workflow, where it is in the process, and you can kind of track its progress. So now it's going to ask me, because we're on to ⁓ step one, which is connect your tools, it's going to ask which platforms am I using regularly in my work? So it's asking, do I use Google Workspace, Notion, Slack, a CRM tool? So just to keep things simple, let's just do Google Workspace for now. Now it's asking, is there a primary tool you work in more than any other ⁓ within the Google ecosystem? So let's just say ⁓ email for Gmail. And then it's going to walk me through the steps of connecting Gmail and my Google Drive and my Google Calendar. Those are actually already connected. because I didn't disconnect them just for the sake of this demo. So it should most likely search through my connectors here, understand that it's already connected and skip to the next step, which is fine because the process of connecting your tools is so simple. All you do is literally log into the tool within Claude and then it's connected. So there's nothing really to show in that regard. So it's saying those three are already authenticated exactly like I said. And then it's going to rerun the verification test prompts that the guide recommends just to confirm that they all work end to end. So it's going to do those three in parallel. I've got to explicitly give it that permission and I'll come back once those tests are done. All right. So it tested to confirm that all the three connectors are working. So it says Google Talender is working. Google Drive is working and Gmail is working as well. And it even specified in a real fresh install workflow, you would have just clicked the three connector buttons in a pop-up. gone through Google's authentication screen three times, and then you would be done. So that's how simple the process is for connecting your tools to Cloud. Now, next, it's gonna have us start building out our context files, which is step three in our progress bar here. So it's asking in one sentence, how would you describe what you do? Let's say in this case, ⁓ creator slash content business, how many active businesses or projects are taking up your time right now? So in this case, just one all in. And it even told us now that we're on capability to context files, it's going to produce three files from this, this particular step, the about me, the brand voice, and the working style, which is exactly what we want it to do. Now it's asking, what is the business called and what does it actually do? Pick the closest. You can add detail and other. So in this case, let's call it a community business called build with AI hosted inside school. where we teach and I'm going to use Whisperflow to talk this out instead of typing it, which is a lot quicker. Solopreneurs and small business owners, how to install an AI operating system inside their business. Now it's asking, what's your professional background before this? So I'm going to, again, talk to text here. I worked for IBM for two years. From there, I owned my own e-commerce business for nine years. And now I run this community and have a content business full-time. All right, so again, it's just asking me these questions to get context on who I am so that it can build the context files appropriately so that once it has those context files, it never forgets anything about me. It doesn't have to ask me these questions every time. It can just refer to those files. So now it's asking, who is your ideal customer inside Build with AI? So I'm going to type in my answer here. Non-technical solopreneurs and small business owners doing anywhere from $500,000 to $5 million per year who feel like they are the bottleneck inside their business and want to scale without hiring additional employees and without needing to learn how to code. Beyond Google workspace, which tools do you live in daily? So I'm going to check the box for school. We also use Notion. I also use social platforms and I also use these video and podcast tools. So I'm going to check all those boxes and see what it has me do next. All right. Now it's done creating the about me file. So it's moving on to our brand voice file, which is trying to get a feel for how I communicate and how I talk so that when Claude coworker creates any content for me or it writes any emails in my voice, for example, It's going to write them exactly the way that I would actually write them or say them. So now it's asking, how would you describe your tone when communicating professionally? Certainly option one direct and no BS get to the point. Do you communicate differently depending on audience? I would say option one, same voice everywhere. Now, this is going to generate kind of a bare bones version of your Brand Voice profile. What you can and probably should do is once you get through this onboarding, you can go back and tell Claude to use some reference files to beef up your Brand Voice profile. Now, all that means is you're giving Claude examples of your work, whether it's content you've written or emails you've sent or proposals you've drafted in your own voice by hand. So that way Claude has even more concrete examples of what you talk or how you talk and what you write it like. So it's saying what kind of language makes you cringe, thinks Cludge should never sound like. So in this case, it's asking me buzzword corporate speak, absolutely over promising hype, gatekeeping jargon, generic sounding AI content, all of the above. And then I'm also going to write in the something else box, never ever use ⁓ dashes. Do you have existing brand materials, Claude should reference? So in this case, it's asking yes. One of the options is yes, search by Google Drive for them. So that's one of the benefits of connecting your Google Drive or your Microsoft SharePoint or wherever all of your documents live, is Claude can actually go into your drive or go into your files and look for those reference files automatically without you having to search for them manually. So in this case, for the sake of the example, I'm just going to say no, just use what we discussed. even though I do already have an existing brand voice skill. And I could easily have Claude search through my Google Drive to find examples of my brand voice, but to keep it simple, I'm going to just have it use the information I already gave it. Now it's asking ⁓ for my working style, right? It's going to ask some questions to build the working style file. So it says, when Claude starts a task for you, what do you prefer? Always ask clarifying questions first. Just go on simple tasks, ask on complex ones, show me a plan first and let me approve. So in this case, I like it to show me a plan first and then let me approve default file format for deliverables. When I don't specify, I'm going to say, ask me each time because usually if I want a Google doc, I will say so. And if I want, for example, a spreadsheet, I will say so. Now it's asking how detailed should Claude's responses be? I would say match the complexity, short for simple and detailed for complex. Any hard rules for how Claude should behave? No fluff or filler, absolutely. Show reasoning when it matters, bias towards action, be opinionated. So I'm going to check all of those boxes. And you can tell that this onboarding is, it's pretty in depth in the sense that 99.99 % of people will never ever do this when using cowork for the first time. So if you simply carve out the 20 to 30 minutes that we're going to spend here, setting Cloud Cowork up right the first way. Every time you use cowork from here on out, it's going to be literally a hundred times more effective than if you hadn't done this. So I highly recommend getting the free plugin in the description or show notes so that you can just do this by yourself in 20 or 30 minutes. So now you see it's saying it's got everything it needs. It's gonna write all three context files to the context folder inside the folder that we originally chose as the destination folder for this project. It says no ⁓ dashes anywhere, I promise. Now in order to create the file, it's going to ask for my permission. So I'm gonna click always allow and I'll be back when it's done. All right, so now that it has created my context files based on those context files, it's actually going to go ahead and proactively move to capability three, which is writing my global instructions. So it essentially looked at those three context files that we created and then wrote a set of global instructions that ties them all together. Essentially, this is what tells Claude when to use which of those context files and what they actually mean at a high level. It's saying that it wrote the global instructions. It was a total of 717 words. And in order to activate it, all we do is we go back into Cowork and we edit the global instructions and literally just paste the contents of the global instructions file that it gave us and then save. So from that point on, every Cowork session is going to load with that context automatically. That's the one key step that most people miss that makes Cowork significantly more effective. So now it's going to move on to capability four, which is step five in our progress bar here, where it's saying it's going to start installing skills. So it's going to search the plugin marketplace to find skill and plugin bundles that match my role. So there's lots of different skills and plugins available in the marketplace. These are skills and plugins that have been created by Anthropic. So they're vetted. They don't have viruses. They don't have malware. And now that Claude knows about me, in my role, in my business, and everything that I'm doing, it's going to go and search that marketplace to find skills that I probably need that would probably be beneficial to me. It's essentially saying I already have a mature skill stack installed because I do already have a lot of skills installed because this is the main place that I use Cowork. If I were you watching this and this was genuinely the first time you were using Cowork, this would be much more effective for you because it would go out and find the skills that you have not yet installed that would be beneficial to you or your role. So it's basically saying it found the document creation skills that would be beneficial docxpowerpointxlpdf. It's saying that I already have a set of content creation skills and it recommended from the marketplace the one gap worth considering, the one plugin that I don't have installed is the marketing plugin, which is an official one from Anthropic that helps you with email campaigns. ⁓ email sequences, content creation, brand review, SEO audits, all in one plugin. So that's one that I probably should go and download. Now, now it's going to move on to schedule tasks, right? Which is step six, capability five in our progress bar. What these are are prompts that run automatically on a schedule inside of cowork. The one thing to keep in mind here is that schedule tasks only run if your computer is on and your cloud desktop app is open. So if your computer goes to sleep or if you close your cowork app, your schedule task will not run. So keep that in mind. Now it's asking me which daily or weekly rhythms would be useful to automate. One that I really like is to have a morning briefing. So essentially weekdays at 8 a.m. this is going to send me a morning briefing, which is probably going to scrape my calendar, my email inbox, my to-do list, all my tools, and essentially give me a rundown on what it is that I have to do that day. So it's also asking if I want to include the quarterly context review, which is a prompt that we built into this, ⁓ cowork onboarding plugins. So if you use our onboarding plugin, it's going to ask you this every time. So I recommend you say yes, because essentially what that's going to do is it's going to run a scheduled task once a quarter that tells you, Hey, update your context files. Because more than likely. your role or your business or things are going to change at least once a quarter, if not more often. So by running that schedule task once a quarter, coworker is going to remind you, update your context files so that Claude has the most accurate and up-to-date information on you and what you're working on. So it's asking if I want to create these two schedule tasks as is. Yes, create both as proposed. So what it's proposing for the morning briefing is that ⁓ Every weekday at 8 a.m., it's going to read my context files first. Then it's going to list all the Google Calendar events for that day along with the times and attendees, and it's going to flag any prep notes attached. Next, it's going to search Gmail for any emails received in the last 24 hours. It's going to flag anything from clients or partners or anything that requires a reply. Then it's going to organize the briefing by urgency based on things that need action today. versus things that are just an FYI. And then lastly, it's gonna save that briefing inside the folder that I chose for this project. And that way I can go back and review that at any time that I like. And then for the quarterly context review, it's saying on the first Monday of January, April, July, and October at 9 a.m., it's gonna read my context files, it's gonna summarize what each file currently says in one paragraph, then it's gonna ask me. Has your business changed? Are you using any new tools? Do you have any new preferences? And have you updated your voice rules? If I say that there are changes, it's gonna update the relevant files for me automatically, and then show me the differences before saving. Then it's gonna check to see if the global instructions need to be regenerated to match. And then it's gonna save a short review to the outputs inside this project folder. So that is the gist of the two schedule tasks that it's proposing for me. And so I'm going to say, yes, create both as proposed. And we're going to let cowork create and schedule both of those tasks and we'll be back when it's done. All right. So now it's telling me that both tasks have been created and it's also telling me to go ahead and run the morning brief now ⁓ manually, because that's going to pre-approve Gmail and calendar access so that future automated runs don't pause waiting for permissions. So that would be smart for me to do run at one time, grant those permissions once, and then it'll run. every other time on autopilot after that. Now it's wanting to run through the security review. Now this is completely optional. However, ⁓ Claude recommends we do it because essentially it's a five minute test that understands what we've done so far and locks in any sensible rules. It's going to ask four interview questions, audit our setup, and then generate a security posture document. So let's go ahead and go through that. It's asking what kind of data does Claude have access to through your connected tools and workspace? It's asking essentially about the sensitivity of the data that Claude has access to. So nothing I have access to is sensitive. I'm just going to say mostly business content, docs, email, and calendar. Next it's asking how cautious should Claude be with external actions, whether that be sending emails, creating calendar events, publishing content, which is funny because Claude cowork can't actually send emails. It can only draft them. But I'm going to choose option one, which is ask before any external action, not just ask based on sensitive actions, ask me for any external action so that way it never does something that I didn't want it to do. And I don't have the excuse to say I didn't give it that permission. So now it's asking for schedule tasks that run automatically. What's your comfort level? Let's say read only tasks should be fine to run automatically. So I'm going to choose that option. How often does Claude process documents from people outside your organization? Rarely, it's mostly my own content. So I'm going to choose that option. All right, now that should be everything that Claude needs to run that analysis and give us back essentially a security posture document. And what that's going to tell us is essentially, hey, here's how much risk you have associated with your Claude cowork setup based on the types of information you're dealing with, whether you're dealing with just internal documents or whether you're dealing with external client documents, whether you're using or interacting with sensitive data such as social security numbers or financial information. If you're not doing anything like that, if you're using Co-Work mainly for internal work processes with your own data, this should be a non-issue. But let's see what Claude has to say. So it is telling us that the workspace is clean. Only the five files we just created are loaded in this project. There are zero sensitive patterns detected. Now it's gonna generate the actual security posture document, which again, shouldn't tell us much more than we already know, but I'll show you that when it's done. Okay, so it wrote us a rather long security posture document, but one thing that it's asking that I absolutely love, and again, this is something that the everyday person simply would never do this because they just wouldn't think to do it, but it's asking if I want to add the security rules to our global instructions so that they load every cowork session. Absolutely yes. Why would we not do this? So essentially it's just making Claude a bit more secure. by adding it to our global instruction. So I'm gonna choose yes there. So now it's asking if I want to run the quick self-assessment, which as I mentioned in the beginning of the video is not something that you guys are gonna have access to inside the free plugin. That's something that only the folks inside of our Build with AI community are gonna have access to inside their very special version of the plugin. And so again, that is going to essentially interview you to audit your workflows and then identify the highest ROI opportunities inside your business that Cloud Co-Work can either automate or take off your plate or you can build skills to get those processes off your plate. So in this case, I'm just going to say, let's conclude the onboarding now, summarize what we accomplished during this session. So we're going to have cowork essentially summarize what we did during the session. And if you guys are interested in the full version of this plugin, which includes that audit and the skills recommendations tailored to your business, you can join our build with AI community, which we will put a link in the description on YouTube or in the show notes. If you're listening on the podcast to check that out and join us inside the community. Now it's going to essentially look at everything that we did. It's going to give us a little summary here at the end. We're going to go through that and then we'll wrap up after that. All right, so now that we've completed the full onboarding, this is everything that we did. So we built our workspace, which includes our context files, our projects, our templates, and our outputs with read only source permissions. We connected and verified Gmail, Drive, and Calendar. We created three context files, which were About Me, Brand, Voice, and Working Style. We generated our global instructions, which ties everything together. which was a total of 747 words that we still need to paste into our settings. Cowork recommended a entry-level skill stack, and then it even cross-referenced those with our existing skills to see which ones we already have so that it didn't recommend any duplicates. We created two scheduled tasks. One was a morning briefing, and then one was a quarterly context review. We completed a security review, and then it generated a posture document. and merge those recommendations into our global instructions, we conducted a self-assessment, or I guess we skipped over the self-assessment because that's not inside the scope of this particular plugin. And then our final delivered outputs are global instructions, our security posture, recommended skill stack, and onboarding checklist. The context files we created are about me, brand voice, and working style. The only two steps left to do that are manual are simply pasting the global instructions document into our actual global instructions and then manually clicking run now on the morning briefing task just once to pre-approve Gmail and calendar access. So if that was helpful for you guys, please subscribe to my channel and like the video or if you're listening on podcast platforms, subscribe to the show because we come out with at least two incredibly tactical videos and podcast episodes like this every single week designed to help you. If you're a 500K to $5 million per year, solopreneur or small business owner, build your AI operating system so that you don't have to hire more employees and you don't have to write a single line of code. So thanks for watching and we'll be back soon.