Claude just introduced a brand new analysis tool that completely changes how it does data analytics in the background. And the way it does it now is Claude can write and run code to process data, run analysis, and produce data visualization in real time. So it's combining this new tool with something they've already had called Claude Artifact. And I've been using it for a little while now, and it is a huge upgrade from the previous model.
And I think that's for two reasons. Claude, just a couple of days ago, introduced a brand new version of Claude 3.5 Sonnet, which was designed for better analytics and better math and better reasoning. So they just combined that with this release to give us a much more improved data analysis tool.
Now, the way this works with this new analysis tool is Claude is going to write and run JavaScript code when you give it a prompt. So you don't even have to worry about the JavaScript code, but it does show you that. And then it will process data, conduct analysis, and produce real-time insights.
This is available right now to all Cloud AI users as a future preview. I'm going to show you how to activate this because by default, it's not activated just yet. But the interesting thing is I have been using Cloud for these type of things, but it was more for visualization.
For accuracy of data, it was not quite there. So I think the way now is analyzing it with JavaScript code. It's... With the new model, with the new Sonnet model is why it's actually a big upgrade.
Previously, a lot of times with financial data, for example, I couldn't really count on the data. ChatGPT has the same issue, right? A lot of times it doesn't give you 100% accurate data. And when it comes to financial analysis, sales funnel analysis, those type of things, you need 100% accuracy or it's not that useful.
And I'm not sure if this gets us to 100%, but so far, every single test that I give it, I... double check the data and he was right so we'll see we'll do some more testing here think of this analysis tool as a built-in code sandbox where claude can do complex math analyze data iterate on the different ideas before sharing an answer again because they just rolled out a brand new sonnet this actually now makes more sense and they kind of release these on the same week now Now here's a quick example. I'm going to show you this with my account right now. But you just upload like a CSV type of data file and it writes code. It analyzes it in the background and it visualizes it on the right side using Artifact.
And here's just a few different examples and I pretty much have an example for all these. But as a marketer you could upload customer interactions and get information across your full funnel. You could give a data about your sales data as a sales team.
Product manager could upload engagement data. Engineers could upload a performance log from servers, financial teams could upload financial documents, balance sheet, P&Ls and get insights and trends, all kinds of different use cases. These are the five they mentioned over here. Let's jump in and start using it. So go to cloud.ai, again this is available to all users but you have to activate it.
So you'll see this little icon right here, if you click it you'll need to turn it on right over here. So by default it's off, you need to turn it on and all your new chats are going to have access to this moving forward. Now on the left settings panel too if you don't see it there under future preview it's gonna bring up this box where you could turn it on here too. Now I'll start with a couple of samples of data I just got from my YouTube channel so I'll just show you how it all works and then we'll jump into marketing sales and some other data.
Okay here's a simple CSV document and this is the age of my viewers for this channel so this is gonna be interesting let's ask it to analyze this. Now I have some prompts that I'm gonna give you to analyze all kinds of different documents but usually it's going to go ahead and analyze it. Now you see it's analyzing it with writing some code.
That is different than what it used to do before. It used to be able to analyze documents but nowhere to this extent with this new Sonnet model is doing a far better job on the output, which is the whole point of these. Now on the right side it's writing some React code here to actually visualize it. That's a whole different thing. This is called artifact over here but it's combining it.
Viewer age distribution and engagement. Okay so 35 to 44 year olds that's the biggest chunk over here with nearly 25 percent. Watch time distribution again 35 to 44 year olds which is my age bracket here at 27 percent.
Okay very interesting that is very useful visualization. Now I have a much more complex document I got From YouTube, that is all the data in the last 30 days of the YouTube channel. So it's impressions, views, subscriptions.
This one had like 9 or 10 different columns here. So it's analyzing it again, wrote some code. You could see the code, by the way, right here. If you wanted to see the code. Oh, wow, that is pretty messy right here.
But if I hover over it, it's telling me in the last 30 days, these are the videos that performed the best. So this is very useful for someone like me, right? 78,000, 77, 73, 68. And let's see what else we got. We got content theme distribution. So chat, GPT, cloud, AI, and Gemini.
Those are the videos I guess I covered. And we got click-through rate. Again, a little bit messy. So it probably requires a follow-up prompt.
But look at this. It also gives you follow-ups on visualizing. So visualize my subscription growth. And then it'll create a prompt.
And then you'll send out that prompt. And it'll do another analysis. Okay, this is interesting. So it's giving me the videos I recently uploaded and how many subscribers I gained from each.
So it looks like this one, 1800. Alright, this is actually really useful. This is not what the graph gave me at first glance. The chart is way more confusing and I couldn't get a key graph insight out of this.
And again, it's giving me more updates. Now let's get to some more practical stuff that's not for YouTubers. Now the next one is for marketing analysis. So I have this document here.
And it just has a bunch of products and the income from each product, how many of each were sold and it's a massive document. So I'm curious to see if we could upload this document. It says long prompts or documents may take multiple minutes to respond. So okay, that's fine.
Let me see what happens if I don't give it a prompt. Let me see by default, does it go ahead and analyze it? Okay, the message will exceed the length limit of this chat.
Try attaching fewer or smaller files or start a new conversation. So this is one of the limitations with cloud your context window and the ability to add large documents like the csv document that's i think over 3 000 lines not possible which is a little bit of problem because for example if you have a huge customer list 3 000 is not that many right and that's going to be a problem if you have a big e-commerce list like this with thousands of products again it's going to be a problem So I just cut this down to 1200 lines here on the row side and it still has ton of columns here. Okay let me go to my prompts book here. I'll include this below that you could download from my website will make it a little bit nicer and you could get it for free and I'm gonna take one of these prompts so it has prompts for analyzing data sets and you just have to say visualize for example analyze and visualize in the beginning and then you could use some of these here and this has other things for data as far as financial data goes. It could also do some other things like analyzing code and things like that.
Okay, it worked this time. So here's the average spending by product category and the formatting of that document actually wasn't into category. So it looks like you put it into category for us.
So wine, fruit, meat is telling us that amount that was spent here. Okay, made us a couple of different graphs, customer spending segments. So high spenders, low spenders. All right, that's useful.
Let's see the key insights right here. Wine represents the highest average spending category. 26% of customers are high spenders.
I mean, extremely useful stuff. Now, if you've been using this artifacts feature, it kind of looks similar, but the data analysis is totally different. That's the whole big upgrade here.
You have a new large language model, a new version of Sonnet, basically, that is designed to do a better job with this and a whole new analysis tool that is analyzing in a whole new way, which then gives you much more accurate data sets, which is the whole point of data analysis using AI. Okay, this time was a huge document, this Bitcoin price chart across eight years with daily prices. And this is Bitcoin's yearly performance analyzed through from 2015 all the way to 2023. 2023 data is partial until July.
Okay, that's useful to know. And again, really useful graph here that we got out of it. Key milestones, notable years.
Best year 2017 up to... 1200%. Wow, worst year 2018 down 74%. Okay, let's try something fun here. Can you make a projection of the price using this historical data for October 24, 2024?
Okay, Bitcoin price as I'm recording this video is $67,494. Let's see what we got. Oh wow, look at this, 69,000 off just a couple of thousand on this moderate case.
But I mean I guess it's a huge range here from $41,000 to $97,000 but kind of interesting. How did it get that? Current price is $30,000 so This was in the middle of 2023 where this ended. Time to target 1.3 years. Recent volatility 2.47%.
Historical context, Bitcoin has shown an average annual growth rate of 141%. Okay, important things to consider. Past performance does not guarantee.
Okay, this is, I guess, a disclaimer. So again, not financial advice. And this is telling us this is not...
Cryptocurrency market are highly volatile. Okay, now the one limitations that I've been pointing out is the context window, the size of these documents that you could upload, conversation is 241% over the limits right here. And I actually have a paid version of Cloud.
I'm using Cloud 3.5 Sonnet, and that is gonna be a limitation. This one, by the way, is 16,000 lines and I think like 30 different columns. So a huge document as far as the size of it goes, but this...
is going to limit us because now I have to kind of chop this up, analyze it, add it up. That's not going to quite work out for what I'm trying to use this for right now. And even if I chop this up into two different files and try to use it in the same chat, it's going to have the exact same problem.
So I just give it the first 4,000 lines and I try to give it 2,000 lines after that at a time. But even the first one, it put me over the limit. So I do have to remove this one here.
Now, within the limit, it's still going to tell me that this is going to take. multiple minutes to respond. In my experience, it's been much faster than that. But this is, again, a 4,000 word CSV document, which is typically a text or CSV document is what you want to give it, not a PDF.
You want to convert those so you could actually analyze those. And when I give it that smaller file, it did give me useful information here. It gave me a bunch of different graphs and useful charts. But here's the problem. If this is only the top 4,000 lines of that document, well...
This might not be accurate for what I was analyzing for. That was a 16,000 line document. So the context window is going to be the limitations right now, even though I think the analysis is really good.
It's not going to solve some people's problem that are dealing with much larger files to analyze. Okay. I'll finish up with one of my favorite cloud features.
And this is one of my YouTube analytics, again, top 10 cities by views in the last 30 days. And London is number one with over 10,000 views. came from London in the last 30 days on this channel and I'm in Chicago actually and that is not even on the list over here so I'm gonna go ahead and press this right here publish I get this link right here and I could copy this link and I could share this with anyone as a web page that shows my entire presentation I've been using this one quite a bit for these type of visual presentation that artifact gives you but now we have a new way to do this with the new data analysis that is much more accurate And I'm also in the process of working on an entire data analysis course and it's heavily going to be focused on what Claude and the ChatGPT data analysis is going to do.
So this is going to be part of our Skill Leap AI platform which is a bundle of courses that you get for one subscription with a free trial. We have about 26, 27 different courses right now and I try to release one every month and we have other instructors too. So I'll link that below too if you're not a member.
There's a whole community section there. where you could ask us direct questions and we give you all kinds of different resources like some of these downloadable pdfs that i sometimes add on youtube thanks again for watching i'll see you on the next one