Transcription
Oh, wow. Love this. How about that? Since I released my Claude Co-work video, the question I've had more than any other is, "Are you still using Notion? And if so, how do you use them together?" Well, this video answers that question. Short answer, yes. And I'm using them together every day. I got to say, the results are absolutely awesome, including presentations like this.
By the end of this video, then you'll know exactly when to reach for each one. Why Notion is where I think teamwork and complex context gets managed and works. And Co-work is where thinking, building, and strategy happen for me.
I'll also show you how both my Notion and my Claude setup have the same architecture: specialist agents, knowledge bases, and skills libraries running in both, and how you can do the same.
Now, I spent the last 6 months trying to work out which AI tools are the right choice for founders, small businesses, entrepreneurs right now without needing to be some technical wizard. The two that keep coming out on top have to be Notion and Claude. And whilst AI development seems to be shipping weekly, here's where my current setup sits: Notion personal agent with specialist modes built for my own day-to-day work; Notion custom agents running in the background for the jobs that earn their keep and offer a return on investment; and then Claude Co-work on the desktop side for thinking, writing, and cross-file work.
Now, I recently released Co-work OS, a templated file system and instruction set that onboards you to Claude Co-work, which really mirrors my business OS setup in Notion. It's built for Claude desktop, and thanks to all the current beta users using it, send me your feedback via the links in the PDFs. I'd love to hear from you. And well, everyone else, please go and check out Claude Co-work OS, link below, because people seem to be loving it, and I am too. The best thing? It actively onboards you better than 99% of users to use Claude Co-work. So, go check it out if you're new here.
If all this is starting to feel like you're the one falling behind, relax for a moment. The truth is that around 0.3% of people are actually using useful bespoke AI right now to get work done. So, you're not late. You're genuinely early, but that window is probably closing. And well, with this video, I'm about to move you further up the leaderboard. So, check you're actually subscribed here, and let's get started.
My Notion setup has three layers. And once you see them, you'll spot the same pattern in everything that I'm doing across my systems. There's the business OS layer. This is essentially the workspace that we've built out that we can interact with. We've got a content manager where we manage our videos. We have areas, uh dashboards, projects dashboards where we can view our projects and track processes. It allows me to look after my team resources, goal setting. There's a meeting notes system where I'm using AI meeting notes to record. You can start transcribing and then use AI notes to summarize. So, all of that is just working in a pretty standard way in Notion.
Now, the key to a good first part of a system is going to be a solid database system. And so, we manage everything in system databases down here. They're all listed in this area. And that just allows me to be able to keep all my information consolidated in one place.
Now, the personal agent is the second layer, and this is the kind of assistant of specialist coaches and tools within Notion that knows Notion better than any other agent. It's the closest thing I've got to a small team. It allows me to build things like a productivity coach, a decision coach, a content strategist. I've even got a business strategist. What's great about this is it's one entry point that picks the right mode based on what I ask it. Whatever I ask it, it will then pick a set of specialist instructions designed to work within Notion and follow them, meaning that I get these specialists without burning through credits. And it allows me to do the same single chat with loads of specific specialists.
Take a look at this. We have an agent hub as part of our system. The global instructions in Notion for personal agent, you can set a single page that will influence its behavior. And what I've done with Agent OS, a templated system that I've built, is create a mode protocol. It essentially selects specialist agents. They each have their own sub-agent playbooks. Now, this will look different to my current templated version because I've been updating it. And what I'm doing is building out sub-agents in this system here. I'm also creating a skills database that backs that up. So, it means that each agent is related to a set of skills.
So, if we go back to the agent hub, this is the latest development I've added to this: below that, a skills and standard operating procedure library. What's brilliant now about Notion, much like in Claude, is you can create skills. And the reason that I want this in here is because Notion's personal agent is an expert on using Notion. The other thing, if you're going to go and use Claude and action via the MCP, take actions in Notion, big jobs like writing entire databases and pages. One, it's not as trained on it, but two, it's going to cost you a lot of tokens to connect via the MCP. Using it within Notion is going to allow you with personal agent to select any of these models: Claude, Gemini, GPT, all of the rest of it. I have pretty unlimited use at the moment of it. So, it's super powerful.
Now, yes, it's only prompt and response, but you can do a huge amount in there. So, I'm just going to test this new Proforma version out. And what's going to happen is when that's running for personal agent, it will follow this mode protocol, pick a specific mode that is most applicable to what we're doing, and run it. And we could say, "Please explain how you operate with your sub-modes and skills." There's an orchestration layer that runs the voice profile, which you can see here: my beliefs and stance, non-negotiables, personality structures. I've written all of this out. And then an anti-AI writing style guide, very similar as you'll see in my Claude system. And then we have all of the modes that it will run, and then there's a skills layer which it can select to use. So, yeah, really cool.
The joy of this is it then means we have all of our information consolidated not only for clear viewing as a team, as individuals, as humans in the process where we can actually see really solid reporting views. We then also have a set of specialists that we can call upon as we work. This allows me to do busy work, strategy, coaching, that kind of stuff with expert agents. And my favorite final addition to the personal agent layer is this: that we could also create agent brains. So, for example, I have an online business marketing knowledge base which essentially just takes loads of key strategies listed into the database that I've seeded, and the agent that is served by it then answers only based on what's in that knowledge base. So, if we go back to online business strategist mode, you'll see it will then be required to follow key instructions, but then also base all of its information on what's in that knowledge base. Really powerful.
So, in Notion, let's select a mode. Let's do Opus because it's really similar. And I'm going to say to it something like this: "Let's make a plan for launching the Aentic Business OS and build on the work we've already done. Please check all of the related notes to the Aentic business big knowledge project that we're building in projects and come back to me with an overview of where we're currently at for launching this and what the next steps might be." So, I've used Whisper Flow to drop that in. I'm going to send that to the agent, and we can let that run.
And while we do that, I can go and check where we're at myself. So, we can go and find our Aentic Business Accelerator plans. And in here, I can see what we've been working on, some of the ideas, where current tasks are at, and I can go and take a look at the new template that we are working on.
Now, what's really cool is you can see here that it's actively selecting mode 7, the online business and marketing strategist. That is the correct mode for this planning process. It's checked out the information. It's had a think about where we're at. It's looked at the actual template system. It's doing a shedload of work that is all very specific to it being in Notion that I just don't think Claude connecting into it would necessarily be able to find as easily. We've worked out our ecosystem. It's looking at pricing, strategic decisions still to make, adjacent work, how we're going to work through the next few months, and I can get this to save it as a report. I can build and work with it.
Now, what I love about this is it kind of becomes your co-pilot. You can work alongside it, check in, ask questions, get opinions, and know that it's basing its feedback and responses on the knowledge base it's linked to. And there is that Aentic business product ecosystem. It's relating into that.
Now, I think what matters most about why I'm using personal agent in Notion like this right now is because I don't switch tools to switch context. I work in one place where our business lives. And for a team, Notion is still the best place to do that, in my opinion. One agent, multiple modes, the closest thing to a small team in your sidebar.
So, layer three is custom agents running in the background. These are the ones with their own page instructions, triggers, and schedules. I use these sparingly. They're credit-based. So, the rule is: high leverage, repetitive, structured data jobs only. An example is something like this: three custom agents that are really earning their keep right now. A customer support playbook curator. This sweeps the support inbox for better creating for templates and sales. It drafts SOPs and playbook responses that have proven to work. This means that we're gradually building a system that would allow an agent to triage first emails. Pretty cool. Things like a weekly briefer that can brief the team on what's coming up. And one of my favorite new features in Notion is the ability to autofill with database agents. So again, these are automated custom agents that work within databases, something Claude couldn't do specifically. I'll show you what that looks like in just a moment.
But first, we need the qualifiers for how and when we would use these. And I'd answer these two questions before the agent earns its credits in Notion. One, would you do this every week if the agent didn't? And does it have a clear, measurable return on investment? If both of these are yes, then it's worth setting it up in Notion and probably not in Claude. I'll talk to you about scheduled tasks, routine tasks, and saving money on automated processes with agents in a little bit when we move on to Claude Co-work.
So, let's take a little look at a custom agent in the wild for both databases and regular tasks. So, if we go into the sidebar here, you'll see my custom agents. So, here is Walt, my weekly briefer, and we have settings up here where we can have him scheduled on a trigger every Monday, and he just sends me an email. He's got a workflow guide. He has tools and access, and he has triggers that run him. The best thing about this is we then also have monthly usage reports for the credits up here. And we can set a limit to the amount of money we're happy to use now, which is really good.
There's also great news on Notion custom agents, which is Notion have slashed the prices. But not only that, they've also given us down the bottom here far more options for which model to use. And for a reporting agent like this, Opus is massively overkill. So, we could just strip it down to a much more affordable version, which is really nice, and it's going to make the whole thing much more usable.
But one of my favorite use cases is Andrea, our customer service analyzer. She runs on a schedule, builds a report from reading our emails and reporting on what's happening. And then the best thing about it is in settings, if we just go and see the connections she has, she is linked to our customer service weekly report database. And that allows her to create reports. So, we can click into the report, and we get a full list of trends, issues identified, successes, some nice quotations that we can then automate and send into our processes. Really cool stuff.
And as for those database agents, if we swing down into our workspace, let's go and check out. So, here's a great example. We've got my productivity and life management knowledge base. This is a knowledge base for my productivity coach that can uh help me with specific frameworks and strategies and guide me. Now, let's say that we wanted the key insight to be updated automatically when we add new items. We can click onto this and use AI autofill. And you'll see we have the option to do either basic or custom agent. If we click on custom agent, we then have database agents which can be used to specifically work within a database. This increases intelligence. This allows you to work out credit usage. Select your agent model. So, we could keep it as a simple one. And you can see intelligence is low, cost is low, speed is high. If we swap it to something like Opus 4.7, we see cost is high and intelligence is high. So, this really allows you to be quite granular in how you build out your system. You can get it to update specific items. So, it could fill out entire databases for you. You can set recurrences, on-page updates, on-page creation. There's a ton of stuff you can do here. And you can set in autofill. So, there really is a lot that happens, and this is why I think it's so important to remember that custom agents in Notion are really worth using for the right use case. If they've got a return on investment, if they're managing detailed data or specific workflows that you manage with your team in the Notion workspace, still the best place to manage it across your team. You can set them going with granular control, hone them in for good value responses, and they'll do a lot. So, really cool.
So, that's where my team's work lives. Structured databases in Notion is where we're getting our work done, and it's why Notion Agent is still really relevant to us. But there is a whole layer of work that doesn't really sit cleanly in a database that's shared, and that is local files on your computer. Think of design exports, PDFs, cross-tool research, messy personal admin, or just your general personal work that you need to do on a computer that you don't want to share in a shared workspace. And that's where Co-work comes in. The way that I use it might surprise you because it's not loose or chaotic. It mirrors my Notion setup almost exactly. For me, if Notion is where teamwork gets done, Co-work is where thinking, building, and strategy happens. It's the same architecture but a different tool for different jobs.
Now, I've written here "power in Notion, close on disk." So, what does that actually mean? Well, it means that my Notion system loosely mirrors Tiago Forte's power method. That is the idea of organizing your work into projects, specific jobs with multiple tasks, areas, consistent continuous parts of your work or life that you need to organize information around resources. So, that might be information, research, documentation, SOPs, and increasing with AI is things like templated pages and skills. And then archives, a place where you can keep archive material in case you need it later. What we're really talking about here is how to organize a second brain. So, this video might help you decide which one you prefer, or it might convince you, like me, to use both together.
So, let's look at this power method in Co-work and how it compares in a moment to Notion. First of all, a lot of you might have spotted something like this in Claude Co-work projects. I'm not actually using it. And the reason is I think that using a folder structure can be far more effective.
So, if we jump to file structure here, Co-work OS, this is the folder that you would point Claude at right here. Within that, I then have a set of files. So, we've got an "about me" folder, which we'll talk about later, but then key to this is an "areas" folder. And within that, you have an MD guide for Claude to be able to read, which talks it through what areas are, broad ongoing parts of life, and what projects are, and then how Claude decides to put things into it. And it also gives it instructions about what each element is. And I've actually got a kind of admin PA area within areas which is designed specifically to help you with tasks, be an assistant, and basically use this as a kind of captain's log personal assistant. There's a plug-in that links to it. Most obviously, within this structure is, for example, if we—these are templated examples—but if we took say finance as an area, you then have an archive which can be where items are moved to when we're not using them, and an outputs folder. And each time I create a project, for example, I was building a slide deck creator skill, it will have a memory MD which drops in bits of memory of the process that you've had on that project, and then a project brief which she's created which clarifies where Claude is in the work. That means as you work on the project, you can pick it back up by just asking it to go to the project, and it will read those project brief files, memory files, and then only create items in the output folder.
If you compare that to Notion, this is designed much more as a dashboard system with a set of databases. Areas, if we look at my Life OS template, are really just filtered views. So, if I go into say "top of mind," you then have views of all tasks, all projects, all goals, notes, references that are related to that area. And this is an advantage of Notion. You can build really powerful connections across different filters. So, a project is a filter, an area is a filter, tasks can be related to a specific project. And what you then have is a separate place for knowledge, a section for adding notes, references, specific information that you might gather around your business. In my personal workspace, we have a notes and meetings view where we collect all of our conversations and relate them to the projects and channels so that when you want to see them, you can see them. And I think Notion is still one of the best places to collect SOPs and specific subject knowledge to then interact with your agents. And so, I suppose this is what you would compare to resources.
So, if you take a look in resources, we then have templated output examples. So, take a little look at this. This is an MD file which guides Co-work to how it should output an invoice for me. So, when we run an agent to manage invoicing, it will follow these instructions, and this is what it should end up looking like. It's just a simple text file. We have the same for like what a project brief template looks like, which you saw earlier.
I suppose the direct comparison to something like my agent specialist modes in here, like my decision-making coach, social media repurposing specialist, all of these things would be plugins. Plugins allow you to build a specialist like my personal assistant. So, what that looks like in the actual system here, where you can see we've got this is the personal assistant that I've built as a plugin, and you can use forward slash commands to get it to do different jobs. So, you know, you can do an end-of-day summary, and it will run it just by typing "end of day summary." But if we then go into customize and we look at plugins and personal assistant, you'll see that each personal assistant has a set of instructions on how it should operate. But it also has key skills that it can run. So, I can just type "/log," and it will help me, and you can then view each of these skills here. So, these are all the direct equivalent to over in Notion: we have our specialist sub-agents with their instructions, and then the skills is a library within Notion that is created and aligns with sub-agents that should use it. What I love about this is I can create skills that work across multiple agents.
So, within my Co-work template, I've actually built a specialist sub-agent builder. And what it will do is build you sub-agents with their own knowledge bases. So, if you recall back when we were speaking about these knowledge bases, you can create that in Co-work. If we go into my guides folder, there is a guide. Let's just open this on that specialist sub-agent. And now the way this works is you will get prompted through building the specialist. It will define the specialist. It will seed knowledge into a knowledge base wherever you've got it from, and it will then give you a means to be able to link that plugin to the knowledge base. It packages it all, and basically you'll then have a plugin that you can use in Co-work that is reading from a knowledge base. It will lead you through the whole process.
So, that's the kind of folder and layout shape. Probably the most important thing in my system and how I'm using Notion and Co-work at the same time is about giving my system a place to start from that it already knows who I am. And that's really about an "about me" section. I have a voice profile, a context map of where information can be found, and a "who am I, what I'm building, what I value" page, so you then don't have to rebrief it.
Now, I've had a lot of questions in the comments on Co-work about why you would do this because it will just burn through credits. I don't think that that is the case because, in all honesty, I do think that there is a certain cost to using something like Claude Co-work. It's worth making sure it reads context before it does something. The difference is I get a much stronger first output rather than having to keep prompting. And often reprompting and readding instructions will add to the costs. So, I think it's really worth doing this, but one thing I would always do is make sure it reads your "who am I" page, and then you could always hold things like voice profile and anti-AI writing guides until it actually writes something for you, which is an interesting approach.
But let's just take a little look at how this compares. If I just ask Claude, "Hi, what do you know about me?" As it runs, what you're going to see is it's first of all read its Claude instructions. It's very likely from there going to start reading its voice profile and "about me" pages. There you go. So, it's read the "about me" page, and it's read the voice profile, and it's read its memory. Yes, that's a lot of things for it to read, but genuinely I would suggest it's really worth doing this because it just makes such a difference on how the system works. So, there you go. It's given me the information, and yeah, that's not really the point. I wanted you to see it read these, and if we click on these, you can then see what we have written about me. So, I have a voice profile, core identity. I've actually gone through and written questions and answered them, so then my voice profile is not only how I speak, but it's also a kind of guide to my key ideas and thinking. An "about me" page, this one lists who I am, what makes me tick, my background, key information, brand voice, and it shows a bunch of information about like how to speak and work, which is cool. And then the memory file is essentially working through and sharing a huge amount of information on what has been done. Now, the idea of this memory file is that there are instructions on how it should work. So, how it should format things and how it should prune things. I actually think that this is getting a little bit long and probably is going to cost quite a lot of money. So, it's worth thinking about giving clear instructions to what global memory is and what project memory is. So, those are the kind of key ideas.
I also love using an anti-AI writing style guide. So, these are the writing rules. If you haven't heard of this before, what I've done is ask Claude to take the entire contents of the wiki page on AI writing style and create itself some instructions to never do any of it. It's really, really effective. So, yeah, worth considering that.
I've also built in Claude a context map, and this actually shows the key databases in Notion where it can find information. It can be quite expensive to use Claude to connect into Notion using the MCP. So, when you jump across, you can use connectors to speak to it, but it can be quite expensive on tokens. So, this will reduce token use and get it to the information quicker when you do want Claude to be able to reference what's in Notion.
In terms of what this looks like in my Notion setup, well, it's already got all of its context built in. Really, the map is there. And in Notion, if we go into settings and connections, you'll also have all of these tools, and it can search through those. So, it does have those options. In Claude, it's slightly different. If we go into settings, connectors, you then have a similar setup. And I do find that Claude's interaction and ability to use these tools is really strong.
And while we're talking about instructions, let's just compare them. We have the main global instructions. And that is where I am keeping context maps. There you go, my key databases linked in my main agent instructions, personal agent instructions. And then we have "about the user." And this is where I'm keeping my voice profile and my anti-AI writing style guide and my "about me" page. So, really similar things. What I actually did was write these in Notion, point Claude at them using the Notion connector, and got it to recreate them in Claude. And I think there's a really easy thing to do. These things are not going to change that much. So, it's really worth getting your "about me" files really solid and syncing them across both systems, meaning that you can then operate locally in Notion when you need to, but also in Claude.
And while we're on instructions, those global instructions that it used are these. This is the key bit of information that you need to get right. Make sure that the first thing that's said in global instructions before every task is "read these files," and you can adapt that to be however much you want it to be. But yeah, it really, really helps. And then I think other key things to do in Claude are being clear about what's read-only. I have an exception about when it wants to update certain things. I've given it naming conventions, and I've given it operating rules. If the brief is unclear, use "ask us a question" to clarify.
Now, we talked earlier about custom agents and that they're based on run limits. So, here you can see a thousand credits is around $10. This is reasonably expensive for two runs. You're talking, for example, a weekly customer service analyzer which will look through my emails on Notion at better creating and and give us a review of our customers' feedback and information. You're talking about $250 a run, which is a significant amount of money. If that's got a return on investment, it's great. And that's why scheduled tasks in Co-work are so fantastic. The only thing you really need for scheduled tasks is to be able to make sure that your computer is always on. And the great thing is you can now also speak to it using Dispatch. So, I was having conversations with it. And this allows me to have a really quick conversation with Claude and be able to speak with it on the go. Something that's much harder to do with Notion in a consistent way. So, scheduled tasks, I think, are a brilliant choice for personal repeatable tasks that you want to run automatically without having to pay a huge amount against your plan or on top of a plan.
Now, I am using a Max plan for my system, but I will say, you know, if you look at your plan, £90 sounds like a lot, but if you think about that as all of the work done, a website built, systems operating, you know, I'm doing a huge amount of work, it genuinely does feel like having a personal assistant for that, which is worth it. But these scheduled tasks would easily run on a Pro plan for just like $20 a month.
This is what a scheduled task looks like, and you just give it instructions really similarly. You just build the instructions. This will currently read information. It will work through a bunch of processes and then give me a report. And actually, what's really cool is I even coded myself this little app which allows me to have a quick bit of feedback and information on my day. Really cool. You just edit it, set the frequency, and away you go.
But more recently, Co-work has also released Live Artifacts. Now, whilst it can be much more useful to actually go into my task list and look in detail at where things are at and work on them, sometimes you just want a quick view. And what I absolutely love about Live Artifacts is that you are able to very quickly build a dashboard. And I can refresh that dashboard. This has basically coded me a schedule. This is a view of what's happening over the day. I've got tasks that are due, 16 overdue tasks, which I should probably go and work on in Notion. Emails that are left unread within my inbox, and Slack information that I should go and check out. Now, whilst this isn't interactive and I can't do stuff on it, I can click and it will do things like take me to the calendar. You basically just ask for what you want, and it will create it for you.
I mean, let's just do a quick example. If we go a new task, new artifact, chat with Claude, it just inputs this in: "explain what live artifacts are." Then, "look at my connectors and ask me a few questions to figure out what kind of live artifact would be most useful for me." While that's running, I'm going to show you one other scheduled task that recently ran. And this is a weekly inbox triage. This thing will look through my inbox, and then it just shows me key bits of information that I need to address. This is filtered by my Gmail tag for stuff that I need to look at, not my team members. And it's really effective. So, I really like this. You get a copy of the briefing as a file that goes into the folder system. And I'm increasingly finding that this is a really much more cost-effective way to have something running on demand.
So, here we go. "What kind of live artifact would you like?" "I'd love a YouTube content pipeline." So, let's go with that. "How would you use it?" "Weekly review." And I'm going to say to it using Whisper Flow, "You should have an N8N flow which can gather YouTube data and share it as a report for both of my channels. Can you look into that for me, please?" "Okay, I've enabled the MCP for the YouTube brief."
Now, one of the reasons I love Claude for running automated workflows is that N8N, which is an automation platform, is pretty much AI native. And it means that Claude can build directly in N8N. So, I jump into my workflows and take a look at this YouTube analytics brief flow. This entire workflow to get analytics from YouTube and draw them as data into a report in GitHub and then be fed into these systems was just built by it. I'm not technical. I don't understand this, remember. And it has built it for me so that it can create a link to execute the workflow for this task. I think it's all quite sandboxed, and I'm finding it to be very, very effective.
So, let's see what we get with this live artifact. If this works, this just shows you how easy and accessible using Claude is to run your systems. Okay, we've got it working. It's gathering the information on the channels and the videos. Very cool. We've got an artifact being built. I'm going to click create. It uses the N connection. It's now pulling it. YouTube brief. All right. Interesting. When I trigger it, I get an error message. So, I don't understand that, but Claude does. So, it's really about prompting leadership, and we get something good. I'm hoping we're going to get a little payoff here. Oh, wow. Love this. How about that? So, we've got a full YouTube view, net subs, views. I mean, it's pretty damn good.
Let's just take a look at what credits we've used today. If we go into settings and usage, and I've used 27% of a Max plan for that session. It's not crazy, but it's a significant thing. But we've been doing a lot of coding and working. Not bad at all, really. But get subscribed to Systems Made Better, where I'll go deep on each of these skills, tools that you can use in both Claude, Notion, and other key AI tools to simplify the technical stuff for you non-technical founders, solopreneurs, and creative people. Nice.
Here's why I think Co-work at the moment is hard to beat for most people getting started with AI. You can create full brand pack assets with Claude Design. And then, of course, you can code and build with Claude Code. My favorite one has got to be my personal brand website that I coded literally in 8 hours over a weekend feeding a baby milk. And more recently, Live Artifacts that allow us to create reports.
So, here's what I think Claude can do that Notion isn't as good at yet. The whole file system being at hand, local PDFs, design, exports, transcripts. It's really great for kind of multi-mode formats and form of outputs, things like PowerPoint presentations. What is amazing about Claude Design is you can build design systems. Let's take a little look at the design system. My colleague L built this with just a little bit of prompting, and we ended up with a remarkable design system. It's built on our new website design. It's kind of giving us an example of what a website might look like. We've got a stylesheet and look at this UI kit, brand colors, spacing, shadow system, buttons, form inputs, icons, logos. I mean, it's properly amazing what you can do in this. I absolutely love it.
And the results? Well, let's take a little look. Once you've built a design system, you can then also go and create slide decks, prototypes, and let's take a look at our partnerships deck that we've been building. This was all built by Claude, coded for us to be able to begin a system to share with partners. Properly amazing. But my favorite has to be the slide deck that we just created for this video. You just prompt it. This is what I'd like. Build it for me. It answered some questions for me. And then we ended up with an animated HTML presentation.
So, that's the honest answer to the question I keep getting. Yes, I'm still using Notion more than ever. Claude is just making it way more useful as a companion. If you're trying to figure out where to start, assess if you want your files and information to be local on your computer or in a shared workspace. And second, if you're already deep in Notion, stay with it and consider the combination. Build the structure first. Add a personal agent with one or two specialist modes. Then bring Co-work in for the work that doesn't fit cleanly into a database. If you want to fast track to a system for both, check out my Co-work and Agent OS along with things like Life OS for Notion or templates that are linked below where I've done 80% of the building and setup for you.
Now, to finish off, I frame my thinking on all of this using my CASA framework: Consolidate, Architect, Systematize, and Activate. I won't go deep on it here because you can watch this video next, but how to use it to build one focus system for your work and business. The short version: for every job you do, you're making a clear choice about what happens and where it happens. Make sure to get subscribed, and I'll see you on the next one.