Transcription
Look, I respect Claude and their commitment to just staying in their lane and really trying to get better at things like coding and copywriting because that's what we're using Claude for when it comes to our day-to-day work. And in this video, I'm going to dive deep into all of the different features that are available inside of Claude, including connectors, including projects, including artifacts, so that you can go from being a beginner to being a complete pro by the end of this video. Then you can also choose if Claude will become a main part of your tech stack as well. So let's just dive straight in here.
Now the most important thing that I always talk about is when you are getting your outputs, especially if you're using it for copyrightiting, you want to make sure you have a certain style for your brand. So that's why I like the ability for us to have styles in this section here where we can easily choose the Mitch asset brand or we could choose the Big Moves brand. So, in order to create your own brand, you can just come in here to create custom style and then you can add in a piece of your writing and it will describe your style itself. So, it'll do an analysis of your writing and then determine that style for you or you can give it the parameters for creating your brand style. Before you do anything, I suggest that's what you do to set up Claude.
Now, if you look at any of the leaderboards, so LM Arena, this one here, you can see that Claude is typically at the top. So Claude Opus 4.1 is number one on the overall leaderboard and you'd see, you know, even Claude 3.5 is at the top for co-pilot. You can see here um they you're not going to find them in text to video or image to video because they just stay in their lane. They don't worry about trying to be everything to everybody. And I kind of respect that about Claude. Up here you can see that they appear uh a little bit further down in search, but they now have search capabilities that were added this year. You're not going to find them in text to image. uh you will find it in vision. So we do have the ability to upload images and it will analyze images but you it does not create images but you'll definitely find claude at the top for things like text and web development pretty often with claude opus 4.1 being launched earlier this year then there's been huge upgrades to a lot of different things that we can now do like aentic task real world coding reasoning larger improvements to their models in the coming weeks after that announcement. They've also brought back memory. So like it has the ability to understand all of the context of that chat. It can now create and edit files as well. It has the ability now for 1 million tokens of context. So this is where CLAB was lacking behind all the models for a long time, but they're starting to catch up on that this year as well. Plus, there's so many other tools. I'm going to dive into all of these different connectors that you maybe don't even know are available at the end of this video as well and tell you about some of the ones that we're using to do things like downloading videos and creating new records in in Air Table and all types of really cool stuff.
The next most important thing when it comes to board is the in my opinion the projects. So I believe projects to be one of the most important features that you can use for these LLM tools. So it gives you the ability to create a whole lot of extra context for your task without having to prompt it every single time you open up a brand new chat. So we have one set up for our AI tools review. So it structures our reviews in the right formats. We have one for N. So we have a whole bunch of our JSON files uploaded into this one. So it gives context about how we like to build our automations and then we can do two things really. We can ask it to help with any debugging and errors that we come up against. But if you have enough data and because Claude is so good at code, then we can actually get it started with some of the automations. So it'll it'll build out some of the automations and and write the JSON code required to import into Nan. We have one here for our YouTube assets. So that gets done in an automation, but we have one set up in here in case the automation goes down. So we have a backup 11lab scripting, metrical payloads, and we've done a lot of development for our metrical integration into our automations. We did that through here. News image, social posts. Um, so we create these social posts every on Instagram. That's now fully automated as well, but we have it back up here. And so many more like I write my newsletters in here each week. So many more different types of things.
So you have in order to create a project, you just create new. And we're just going to call this test project. You can then describe what this project is for that. Then you want to add some detailed instructions for what you want this project to do and how you want it to um reference any of the files. So if you have five different files in here, so the files are essentially the knowledge base for your projects. So in our instance when we were creating the let's go with the AI tools review we would upload a bunch of files for all the different sections of the review of like how we want it written how we want it structured and then we ask in the instructions okay now you need to reference each of these sections based on files and then create us or help us structure this post for this tool and then it would go ahead and create it or when it comes to the nan project that we set up we uploaded into our files as many as we could of the JSON files that we have for old automations that we've built and then the instructions are your task is to help us both debug and fix errors plus build automations based on our JSON files. So we want to be using those types of things in inside of projects in order to get the models to understand a much larger context of what you're trying to achieve.
Now remember that the purpose of these projects are to allow you to just move so much faster. So, one thing that we do at Make Big Moves, which is our agency, is we help businesses set up projects for each of their team members because you don't always want to be using automations. You you often want to be chatting and using these dashboards. There's a lot of benefit to using these dashboards. And if you can set up projects for all of the different internal projects that you have, all of the different strategies that you have, make sure it has all of the relevant context for all of your team members, then that's how you can get your team to be so much faster as well. So, super important part of using these AI tools.
Now, the other part of Claude that you might not be aware of is that they do have an app that you can download onto your desktop. And when you do that, you have a whole lot more features that become very interesting. So, if I come down to the bottom here and we go to settings, we now have the ability to look at connectors. I mean connectors are available as well inside of the cloud, but you can see here down below before as soon as MCPs came out, we integrated a whole bunch of them inside of our our desktop app so that we could connect to air table so that we could connect to cloudinary which is a storage asset asset and media storage uh cloud plus a few others down here where we can like download videos and stuff. Now, Claude have done a lot to make sure as much of those are available natively for you now as well. And they make it super easy to integrate. So, if you go to your connectors here, you can see a lot of the other softwares and tools that you're already using is now easily integrated into Claude. So, you could add in your Gmail, your Google calendar, your Google Drive, and then be able to chat with any of those from inside of your cord dashboard. other things like inter intercom become very interesting because you can then handle your customer support from inside claw dashboard and then if you come across into into the desktop version so that's why I'm mentioning the desktop app then there's a whole bunch of extra ones down here where they make air table very easily to integrate we've got clickup here as well uh we've got stripe we can do things with uh a whole bunch of extra different apps when using claude so you could read and send iMessage messages directly from Claude as well. So, that's what I like about Claude.
Now, let me give you a demo of what we can actually do with this. So, if I come back into a chat and let's just go new chat and I'm just going to say download this video and let's grab a video from Instagram. So, here. So, let's go onto our big moves Instagram and we're just going to grab this latest video that we have here. And I'm just going to paste that in right here. And so, I've not told it to reference the MCP. I've not told it to do anything besides download this video. And if we open up this, we can see that it's now referencing an integration that we have for downloading videos. And you can say see now, perfect, that it's downloaded this 406 video. It's in this location here directly on my desktop. So if I open up this and I open up the most recent download, then we can see that it's been downloaded right here. So those types of things make it very easy then to I mean this is just one example right and there's so many other examples that we have where you can just integrate different platforms and different tools into the clawed app and which makes it so much easier to work like we the way that we use the download component is we would download it and then move those videos into other social media posts like if we're creating carousels for news announcements and those types of things. So it there's so many different things that you can do when you really start to understand MCPS. If you want me to dive further into MCPS, comment down below so that I know that this video could actually be valuable to create as well and we can go deep into like a whole host of those integrations and how to use them.
The next component is artifacts. Now, we don't really use this at all. I don't see a huge amount of benefit for it just yet. I did try and build an artifact just now and some of these are kind of cool, but then there's no not really a whole lot of benefit to these tools at the moment. One thing that could be nice is like a visual dashboard if you need it for maybe your P&L or something like that that could be built in artifacts. But if we come into one that I tried to build before, I was just querying it on what would creators want and they said maybe a YouTube creator management platform. Now, if we open this up, it's going to look impressive, but it's there's no real value into what it's created because you'd have to start again if you were to develop it anyway. So, let's wait for the load. So, it says that this artifact is completed, but it's just not loading. It did load earlier and it had uh a bit of a dashboard that kind of looked like an app, but there's no real usability in there. I tried even publishing it and putting it to a public URL. It's still not working. So, again, like these artifacts for real business use cases are just not there yet. There's no point in even trying. You could probably get a lot further using tools that are built for this, like Bolt, like Replet, those types of tools to build apps. And even the Google AI Studio ones are way better than what you'll find inside of Claude as well.
Now, the one thing that frustrates me with Claude is that these types of errors seem to pop up a lot. So, I've been trying to get it to analyze a spreadsheet and give me some data. So, I tried first with an XLSX format. I wasn't surprised that this one didn't work because that often happens. For some reason, every tool just hates XLSX. So, I'm not sure why they make it so hard. But then I took the same table and I uploaded a CSV file and it tried to analyze it again, couldn't. So, I gave it a different one to see if it was just that one file and it tried again and still couldn't analyze that particular file. And I see that happen time and time again with Claude is that the analysis component, the analysis tools are very inconsistent. So that's probably the biggest thing that I don't like about Claude. However, we would use chat GPT for any of these types of use cases. So, they do have improvements to make, but overall I really enjoy using Claude almost on a day-to-day basis where we are predominantly using it to write copy, but then occasionally using it to analyze any bugs and errors and help us write some code or JSON files for our AI automation.
Now, if there was one wish I had, I wish that they allowed us to have way more usage for Claude. If you look at the pricing of Claude, it's usually five times more expensive than most other models. So that is the thing that I think they should try and improve and I hope they do improve that because it is such a great tool to use on a daily basis in your business. So now over to you. What do you think? Comment down below. Tell me what I missed.