ChatGPT vs. Gemini vs. Claude -- 6 AI Models in 1 Tool (ChatPlayground Review)

Summary

In this review, Dave Swift explores ChatPlayground AI, a tool that allows users to experience up to six large language models (LLMs) simultaneously. Designed to enhance user experience with a handy Chrome extension, this tool offers flexible LLM access across the web. Throughout the review, Dave shares his user experience insights, discusses the tool's features, provides a walkthrough of its functionalities, and concludes with potential benefits for different types of users.

Highlights

• Discover how ChatPlayground integrates six major LLMs, including ChatGPT and Claude. 🤓
• Learn about the pros and cons of the required Chrome extension for AI Playground. 📉📈
• Experience superior LLM interaction with features like web co-pilot and custom prompts. 🚀
• Evaluate the effectiveness of AI models through side-by-side testing using the 'Strawberry Prompt.' 🍓
• Explore diverse uses, from document editing to image generation, within one platform. 📒🖌️

Key Takeaways

• ChatPlayground AI integrates six LLMs for enhanced user flexibility all at once! 💡
• The tool requires a Chrome extension, which some users might find limiting. 🌐
• Custom prompts and multiple LLMs allow for tailored outputs. 🎯
• ChatPlayground's mixture feature combines inputs from different LLMs for optimal results. 🌀
• Great for quick access to LLMs with the web co-pilot feature, ensuring AI assistance is always on hand. 🤖

Overview

ChatPlayground AI is a versatile tool that brings together six top language learning models, offering users a robust platform for engaging with AI in creative and efficient ways. Users can easily toggle between or compare models like ChatGPT, Claude, Gemini, and more, enhancing the interactive experience.

By requiring a Chrome extension, ChatPlayground ensures users have seamless access to its features wherever they browse, albeit with some software restrictions. Despite this limitation, the accessibility and integration with web browsing make it an indispensable tool for professionals and tech enthusiasts.

From crafting articles and generating creative content to simplifying complex tasks on the fly, ChatPlayground AI proves itself as a valuable companion for anyone looking to leverage the power of AI directly from their web browser. Its functionality promises not just efficiency but also innovation in AI-assisted projects.

Chapters

• 00:00 - 00:30: Introduction to Chat Playground AI The introduction chapter discusses the Chat Playground AI, a tool that allows users to operate up to six language models simultaneously. It includes a Chrome extension for easy access while browsing. The chapter covers an overview of the tool, showcasing its functionality, advantages, and disadvantages, presented by Dave from client.com.
• 00:30 - 01:00: Disclaimer and Getting Started The chapter titled 'Disclaimer and Getting Started' begins with the narrator clarifying that the content is not sponsored, ensuring viewers that the feedback provided is honest. The narrator mentions an affiliate link that supports the channel, leading to a platform called AppSumo, known for lifetime deals. The chapter sets the stage for exploring a tool called Chat Playground, with a promise to review its plans and pricing on AppSumo later in the video. The narrator emphasizes a fair and balanced approach to the review.
• 01:00 - 01:30: Login Process and Interface Overview The chapter discusses the login process and interface for the AI playground, specifically highlighting the requirement to use the Chrome browser and its necessary extension to access the application. This reliance on Chrome and its extension is viewed as a limitation by the author. Additionally, the chapter points out the absence of a traditional username and password authentication system, instead using magic links for login, which is not favored by the author.
• 01:30 - 02:00: Testing Models with a Prompt The chapter discusses the centrality of Chrome and introduces a tool that becomes more user-friendly after logging in. It emphasizes the playground feature within the tool, which, although not the most used feature, serves as the starting point. The chapter then describes a dual-view setup using Chat GPT-4.
• 02:00 - 02:30: Expanding to Multiple Models This chapter discusses using multiple AI models simultaneously, specifically Gemini 1.5 Pro and potentially others, to determine which provides the best output for a given task. The process involves engaging both models at the same time while having the option to turn off web access if needed. An example prompt involved counting the number of the letter 'R' in the word 'strawberry', with ChatGPT successfully providing the correct answer.
• 02:30 - 03:00: Chat History and Sharing Features This chapter discusses the chat history and sharing features of Gemini Pro, comparing its accuracy to chat JPT. Gemini Pro gave an incorrect answer compared to Chat JPT in an example scenario. The text also highlights the ability to use multiple models simultaneously, implying a system of prompts that allows users to toggle between different model responses. The chapter concludes with a demonstration of a split view feature that supports up to four different models interacting simultaneously, including Claude Sonet.
• 03:00 - 03:30: Using and Creating Custom Prompts The chapter discusses the use of multiple language models and the customization of prompts. It mentions the availability of various models including Perplexity, LLaMA, Minstral, and Bing Co-Pilot. The transcript notes that more than four models can be used at a time, possibly up to six, for tasks requiring multiple language models. It also mentions the use of a 'strawberry prompt' and the ability to clear chat history on platforms like ChatGPT and Gemini Pro.
• 03:30 - 04:00: Generating Articles from Transcriptions In this chapter, the focus is on generating articles from transcriptions using multiple AI models simultaneously. The author describes their experience with a particular plan from AppSumo, which costs $59 monthly and allows for 300 chats. They explain that currently, they are using 13 of these allotted chats, and the chapter discusses the efficiency and cost of using such tools to process transcriptions into articles. Specifically, the chapter notes the use of credits when sending prompts to six different models at once, emphasizing the scalability and potential cost-effectiveness of the chosen plan.
• 04:00 - 04:30: Voice Reading Feature and AI Vision In this chapter, the discussion revolves around the usage of multiple models in a voice reading feature and AI vision context. It illustrates how credits are consumed, with each model engagement counting as one credit. The chapter further delves into a specific 'strawberry test' conducted across different AI models, which include Chat GPT, Gemini Pro, Claude, Perplexity, Llama, and Minstrel. Most of these models, except Minstrel, accurately passed the test, highlighting the varying efficacies of different AI systems in processing and recognizing data.
• 04:30 - 05:00: Web Co-pilot: Using AI Across the Web In this chapter titled 'Web Co-pilot: Using AI Across the Web,' the focus is on how users can utilize AI across the web. The transcript provides insights into features such as viewing chat history, sharing conversations through markdown or a shared GPT link, and using multiple models simultaneously. It also mentions a prompting section that is accessible through secondary navigation to enhance user interaction and experience.
• 05:00 - 05:30: Web Co-pilot: Customizing Prompts The chapter titled 'Web Co-pilot: Customizing Prompts' discusses the use of starter prompts that users can explore, particularly for social media. However, the most notable feature addressed is the ability for users to create and save their own prompts. Users can add a prompt by providing a title, description, and category before writing the actual prompt in a designated area. These saved prompts can then be utilized for future tasks.
• 05:30 - 06:00: Practical Use on Social Media This chapter focuses on the practical application of utilizing prompts on social media. It illustrates how to search and edit prompts within an application, specifically under the 'explore' section. The discussion follows the ease of creating and editing prompts, demonstrated with an example called 'show notes'. The chapter highlights the flexibility in adjusting the tone, style, and language of the prompts, emphasizing the transformation of formal content to a more friendly tone. The overarching goal is to craft an article derived from transcription.
• 06:00 - 06:30: Screenshot Tool in Web Co-pilot The chapter titled 'Screenshot Tool in Web Co-pilot' describes the process of using a specific tool within the Web Co-pilot to manage and work with transcriptions. The speaker discusses their intention to use the tool to generate an article, but instead decides to scale back their approach and select just two language models, ChatGPT and Claude, for the task. They clear the chat window and input a lengthy transcription from a recent video along with chapter markers, which are necessary for their custom prompt. The process concludes with submitting the prompt without any further edits.
• 06:30 - 07:00: Engaging with Individual AI Models The chapter explores tools for engaging with individual AI models, specifically mentioning the use of custom prompts to interact with different AI models such as ChatGPT and Claude simultaneously. It draws a comparison with a tool called 'video to blog,' highlighting its ability to capture screenshots from videos and include them in generated articles, a feature not available in the current functionality discussed. The focus is on how these tools can enhance content creation workflows by automating and optimizing tasks like article generation.
• 07:00 - 07:30: Document Editor and Limitations The chapter titled 'Document Editor and Limitations' explores the use of document editing features using two different articles as examples. It discusses a taco truck roundup and show notes for a CAD insert chat, analyzing the effectiveness of different tools used in the video addressed in the chapter. Both tools provided good results with the ability for further tweaking. Additionally, Claude Sonet is highlighted for its outstanding performance, having provided more extensive results, including an FAQ section. The chapter wraps up positively, expressing satisfaction with the results and their applicability for creating show notes for podcasts or longer YouTube videos.
• 07:30 - 08:00: Testing AI Document Suggestions The chapter discusses a feature in a video that allows the user to have the article read aloud. While the voice is described as robotic and not very pleasant, it can be helpful for those who need auditory assistance. For others, it offers convenience, such as listening with headphones while on a treadmill. Additionally, there's a copy function to select preferred outputs.
• 08:00 - 08:30: Template Features in Document Editor The chapter discusses the Template Features in Document Editors. It begins by explaining an overlooked aspect called 'Vision' found in the editor's interface. Represented by an image icon, this feature is not meant for image creation, as commonly presumed. Instead, it allows users to upload images and receive feedback from a large language model based on the visual content. Despite having a basic image generation function included, the chapter notes that its quality is relatively low, possibly due to the underlying technology in use.
• 08:30 - 09:00: Creating Flashcards from PDFs The narrator discusses their experiment with a stable diffusion model, expressing some disappointment. They then attempt to test the vision capabilities of certain tools by uploading an image of tacos and prompting the tools to identify the ingredients. The narrator mentions using chat GBT and Gemini, and notes an issue with uploading to Bing.
• 09:00 - 09:30: Chat with YouTube Videos The chapter titled 'Chat with YouTube Videos' discusses a comparison between two AI models, ChatGPT and Gemini Pro, and their capabilities in identifying ingredients for a taco. It highlights how both models list similar ingredients like taco shells, ground beef, lettuce, tomato, and shredded cheese, but ChatGPT provides a slightly more detailed description by including diced tomatoes and shredded lettuce. The discussion indicates a preference towards ChatGPT's result in this specific instance, acknowledging its detailed output. The chapter concludes by acknowledging the comparative effectiveness of both models and transitions to a new aspect of the platform.
• 09:30 - 10:00: Interacting with PDFs The chapter titled 'Interacting with PDFs' discusses the use of a web co-pilot, specifically a Chrome extension designed to enhance user interaction. Users can set up a shortcut command, which by default is 'command B', to invoke the tool. Additionally, there is an icon available in the lower right-hand corner for easy access. The chapter also highlights customization options for the extension settings, allowing users to tailor it to their preferences. An example provided demonstrates using a Wikipedia article and a chat assistant feature, indicating that this tool aids in efficiently managing and interacting with digital content.
• 10:00 - 10:30: AI Notes and Image Generation The chapter titled 'AI Notes and Image Generation' discusses how to access and utilize content on a page using specific icons and tooltips. By clicking a book icon, users can allow an assistant to access and engage with the page content. The chapter also highlights customizable prompts, such as summarizing page content, that enhance user interaction with the text.
• 10:30 - 11:00: Mixture AI: Combining Model Outputs The chapter discusses a command to simplify summarizing text by reducing it to less than half its original length using a bulleted list format. The narrator demonstrates modifying a command to summarize a page to under 200 words. They copy the command with the intent of making it a permanent change but first want to review the result. The resulting bulleted list provides a helpful condensed version of the longer article, extracting essential information efficiently.
• 11:00 - 11:30: Plans, Pricing, and Final Thoughts The chapter discusses modifying prompts in a chat playground interface, specifically how to change a prompt to have it automatically default to a new one when reading articles. This involves editing the existing prompt under the web co-pilot section and setting it to the desired new default. It briefly touches on the feeling of the interface.
• 11:30 - 12:00: Conclusion and Bonuses The chapter titled 'Conclusion and Bonuses' begins with an explanation of a user interface. The speaker highlights top-level items referred to as categories, such as review selection, edit selection, and reply. The speaker points out that these categories appear prominently on the interface. These main categories can further include sub-items, which may consist of additional categories such as change tone. The focus is on clarifying the structure and navigation of the interface, emphasizing the hierarchy of categories and sub-items within the system.

ChatGPT vs. Gemini vs. Claude -- 6 AI Models in 1 Tool (ChatPlayground Review) Transcription

• 00:00 - 00:30 coming up today on that LTD life I'm going to be showing you chat playground AI this is a large language model conglomerates you to use up to six llms at the exact same time and there's also a handy Chrome extension so you can roam around the web and have access to your favorite llm at the same time it's actually a pretty impressive tool I've been playing around with it already I'm going to show you from top to bottom exactly how it works I'll show you the pros as well as the cons by the way I am Dave from client.com if you've never
• 00:30 - 01:00 seen my content before this is not a sponsored video I'm going to go ahead and just give you my honest feedback on what I think if you want to support the channel you can check out my link down below that'll get you a copy of chat playground or anything on appsumo which is my favorite Marketplace to buy lifetime deals all right so let's get into it I will head over to appsumo at the end of this video and we'll go through the plans and pricing then but for now let's get started right away with chat playground I don't want to be too much of a hater right out of the game
• 01:00 - 01:30 but I am not a fan of the fact that you have to use Chrome in order to use AI playground and you can't even use the app until you install the Chrome extension the login sequence will take you to this page over here which just shows you that you need to turn on the Chrome extension to even access the tool at all if I wanted to be super nitpicky I would also complain about the fact that there isn't a username and password system instead they rely on magic links which are not my favorite thing in the world but I've already complained about
• 01:30 - 02:00 the Chrome centricity of it so we will move on into the tool now the good news is once you get logged in everything gets a little bit better I really like this interface there's actually quite a few AI tools bundled in but as the name implies the playground is kind of the main one it might not be the one you use the most but it's where they start you off and it's obviously named after that so let's look at the playground I'm inside of it right now I've got this kind of two up view here where I can see I'm using chat gbt 4 on the left and I'm
• 02:00 - 02:30 using Gemini 1.5 Pro on the right I've also toggled on web access I can turn that off for either model here and then I can engage with the model it's going to actually talk to both of them at the same time and then I can determine which model has the best output so let me go ahead and just try it out here I've got a prompt how many of the letter R are in the word strawberry make sure to count them all let's go ahead and submit this it's you can see both of the chats in progress here chat GPT gets this correct
• 02:30 - 03:00 and Gemini Pro gets this incorrect so chbt says three and Gemini Pro says two obviously chat jpt in this instance would be the winner I'd want to go with that answer now it gets cooler than this because we can use more than just two models and there's an entire prompting system that I'll walk you through let's go ahead and just check out the other options down here you can see this is the kind of split view I'm in right now if I can go up to three if I'd like and now I've got Claude Sonet joining the party I can even do a quad box and and
• 03:00 - 03:30 have four llms here is perplexity now obviously I can switch this between any of the available models so I can you know have the exact setup that I want the other models by the way are going to be llama minstral and Bing co-pilot so why stop at 4 we've got so many models available we can go all the way up to six models at one time so let's try this strawberry prompt one more time I'm going to clear off my chat on chat GPT as well as Gemini Pro there's a little icon up here which will erase the chat
• 03:30 - 04:00 and now I'm just going to paste in that exact same prompt we're going to send it to all of the models at once notice I'm currently using 13 of my 300 allotted chats that's going to reoccur monthly with appsumo no additional costs here there are higher plans I'm just on the base level $59 I believe plan so uh let's see how many credits gets used when I chat with six models at once it's going to be six right
• 04:00 - 04:30 so yeah you can see that going up it went from 13 all the way up to 19 so every time you engage with multiple models it's going to count one credit per model and just checking this strawberry test here it looks like chat PT got this Gemini Pro got it this time uh it says Claude got it correct as well perplexity also has it correct uh that's llama has it correct and finally Minstrel actually got it wrong there's only two according to Minal the other
• 04:30 - 05:00 icons you're seeing here are going to be your chat history you can click on that and see your entire chat history and there's a sharing icon as well if you click that you can share the chat as markdown or as a share GPT link clicking on this will upload the conversation to share gp.com and then generate a link to share it publicly so that's the playground section in terms of using multiple models at once but as I mentioned there's an entire prompting section it lives kind of coily down here in this secondary button if you click on
• 05:00 - 05:30 this there's a bunch of starter prompts that you can use uh categorized up here so if you want to find some social media prompts you can click over and see what other people are doing there's not a ton of them to be honest but uh there's enough that you might want to check it out what's most interesting to me is the fact that I can create and save my own prompts so over here I can choose add prompt give it a title a description choose a category and then you just write your prompt inside of the body area here save it and then it's going to
• 05:30 - 06:00 live everywhere so it'll be under my prompts here it'll be under the explore section so you can search for it uh very very easy to do I actually created one right here this one show notes I created we can use this I'll click use and you can see now that my prompt says show notes over here and I can go ahead and edit this if I wanted to although I'm happy with it I can change the tone the style and the language so rather than doing formal I will just do friendly and the point of this prompt is to create an article based on a transcription so let
• 06:00 - 06:30 me show you how this works now I don't want to generate an article in six llm so I'm just going to scale this back to two and why don't I choose chat gbt and Claud there we go I'm going to clear out this chat and now I've got a transcription from a recent video I'm just going to paste it in the window here I'm also going to include the chapter markers because that's important for my custom prompt and now that everything is in here and this is long let me tell you like it's it's a long transcription I'm going to go ahead and submit The Prompt with no other
• 06:30 - 07:00 directions because remember I've got my custom prompt with its own instructions up here I'll hit send and now you can see it went to both chat gbt as well as Claude and now the article is being created now if you saw my recent video on video to blog this is a very similar thing that that tool does but I can do it right inside of chat playground the one thing that video to blog does better is that it actually takes screenshots of the video and includes them in the article so that is a pretty killer
• 07:00 - 07:30 feature but here we go we got two pretty good looking articles here you can see this is a taco truck roundup show notes and I've got show notes for a cad insert chat it went through each of the tools in the video and then ended with a conclusion and Claude Sonet did just as good of a job if not a little bit more extensive it even included an FAQ section at the end so yeah very happy with both of those results obviously I could spend a little bit more time tweaking them but I'm in a really good place to start with something like show notes for a podcast or a longer YouTube
• 07:30 - 08:00 video it can read the article to you as well so you just click this little speaker icon and it will speak not in a very pleasing manner it's a really robotic sound but it can actually read the article to you if you you know need it to uh for some people that is critical that they need it to uh but for everyone else if you just want to throw on your headphones and you know walk on the treadmill you can hear the outputs here as well taco truck obviously there's a copy button over here so you can choose which output you like best
• 08:00 - 08:30 and then use it in your desired intended place now the one thing I have not mentioned inside of playground is that there is Vision available as well if you go over to this last icon here which looks like an image this is not for image creation as I originally thought it's actually used to upload your own images so that you can show it to the large language model and it can give you some feedback based on what it sees there is image generation built into this tool you can see it down here as a secondary tool it's pretty low quality I think it's using you know kind of a
• 08:30 - 09:00 stable diffusion type of model have not been super impressed with it but we will take a look a little bit later on I am going to try out the vision capabilities I've uploaded an image of some tacos here and I'm going to ask what ingredients are included in the tacos all right here's my prompt what ingredients are included in these tacos name only the ingredients you're sure of and I'll submit this it's going to go to chat gbt Bing as well as Gemini look like it failed to upload over to Bing for some reason but uh I've got a pretty
• 09:00 - 09:30 good result here from Chachi PT as well as Gemini it says hard taco shells shredded lettuce diced tomatoes shredded cheese and cooked meat mixture likely ground beef with taco seasoning and that is a pretty good assessment Gemini Pro says taco shells ground beef lettuce tomato and shredded cheese so I think in this instance I'd probably give the lead over to cat GPT but Gemini Pro's result was pretty good as well so that is the playground next let's move on to what is probably the more powerful aspect of
• 09:30 - 10:00 this tool the web co-pilot which is essentially the Chrome extension now you can set up a shortcut to invoke this but there's also going to be a little icon in the lower right hand corner by default the shortcut is set to command B you can click right here it'll take you into the screen of your extensions where you can customize it to be anything you want so here is a Wikipedia article in fact I use this in a recent video I've got my little chat playground icon down here I can click on it and here is a chat assistant I want the chat assistant
• 10:00 - 10:30 to be able to access the content on this page so I'll click this little book icon over here you can see the tool tip when I hover over it it says use page content when I click that it lights up now I can engage with the assistant and it will have access to the content that's on the page but there's more there are little prompts that you can use if I go over to the next icon here there's a set of default prompts but I can customize this to have any prompts that I want so common examples would be summarize the page page I click that it just fills out
• 10:30 - 11:00 a very simple command to summarize the page in less than half of the original length let's say I want to modify that I could do it one time right here in the window I'll say less than 200 words and into a bulleted list I'm actually going to copy this because I want to change this permanently I think but before I make that permanent change I want to see what the result looks like all right so here we go this is what the bulleted list looks like it's pretty helpful and definitely cuts through this longer article helping me just get the important bits right out in
• 11:00 - 11:30 front so if I wanted to change this prompt so that every time I'm reading an article I have that exact same prompt with a single click well I'll go back over to the chat playground interface and then we're under web co-pilot and the prompt that I used was over here on the right and it was called summarize so I can just edit this here's the prompt and I can paste in my change prompt and now it will automatically default to that whenever I select summarize now the interface here could feel a little bit
• 11:30 - 12:00 overwhelming so let me just explain what you're seeing the top level items like review selection edit selection and reply these are all called categories and if we keep those in mind remember review selection edit selection and reply then we go look at our interface over here you can see it says review selection edit selection and reply those are the categories that live at the top and then we can have sub items underneath now those sub items can be additional categories like change tone
• 12:00 - 12:30 change length and change structure which then automatically creates a submenu where we have our prompts finally living but if there's no submenu then it just goes right into a prompt let me go back to the other screen and I'll show you what I mean okay so remember summarize key takeaways and questions those were prompts and I can confirm that by clicking on edit you can say see it says edit prompt but change tone change length change structure those are all categories because there's other items
• 12:30 - 13:00 nested inside of them that way you can just organize your prompts to be able to find things quickly and efficiently it's really just an organization system I think the user interface could be improved if they just used a color coding system to indicate which items were categories and which items were prompts uh right now there isn't a visual cue at all to distinguish between the two you just kind of have to assume if there's a nested item the item before it is going to be a category now one of the coolest things about the web Copilot is that it literally is with you all
• 13:00 - 13:30 over the web so let's say I'm on social media here and I want to respond to this tweet well I can simply just select it and then I get the chat playground AI notification I click on this and I have all of my prompts here so I could reply in a positive tone and here is my reply generating right over here I can easily copy it paste in my reply and submit now keep in mind that I'm just using the built-in prompt to to reply in a
• 13:30 - 14:00 positive tone I could get this to be much more specific if it's about a certain topic I could feed it information about that topic in a manner that I'd like to reply my preferences things like that so there's a lot of power baked into having this at your fingertips by the way in the web co-pilot you can change the model I've been using chat GPT but just click this little icon over here and you can choose any of the available models if you have a preference one over the other you can choose that and then use it all over the web last feature I want to show you inside of web co-pilot is Vision we can
• 14:00 - 14:30 actually take screenshots and then immediately submit them to our favorite llm like for example here is the screenshot tool I'm just going to grab a picture of this car right here and I will say tell me about this car I'll submit it and let's see if it can identify ah it can this image is a Tesla Roadster this is the original roster from you know about 13 years ago so very cool it was able to identify it from that screenshot all right so you've seen playground you've seen web co-pilot
• 14:30 - 15:00 there's this section down here that says AI models there's not really a lot to talk about here it's just if you want to engage with just one model at a time you can do so just pick the model you want to talk to and you can engage right there you can also see your chat history inside of this window I can close it open it up again with this icon over there but chat playground is far from done there are a ton of other very cool tools there's a document editor this isn't my favorite to be honest I don't like this one as much but but essentially it's just a document editor
• 15:00 - 15:30 that has a built-in llm that you can have write content for you it's kind of like Jasper it's basically Jasper bundled in I'm going to make a new article here called the history of software I want to publish this on my blog and you can see it is named my document up here and I can just start writing like a regular human being you might be writing in another application you want to bring it into this application then you've got a chatbot on the le- hand side that can see what you're working on and improve your text
• 15:30 - 16:00 so I've got my article that I've been working on over here I can go ahead and just you know edit it as normal but if I select any of the text I get the little contextual menu that pops up to allow me to do things like capitalize improve shorten and expand but probably the more powerful thing is I can just ask the llm in the other column over here to help me improve the article I'll just ask it how can I make this article better and that was the output uh so it's not perfect still some bugs here I've got definitely
• 16:00 - 16:30 some errors going on at the moment let's reload we'll try again give me some ideas to improve this article and there we go I've got 12 ideas to improve the article I could actually have it write some copy and paste it in as well there's you know kind of the standard document AI assistant tools are all available to us here so we are in the chat section of the document editor over here there's also a template section this is very much like Jasper where you've got a bunch of different use cases that you can choose from most of them are fairly
• 16:30 - 17:00 related to business things like newsletters product descriptions SEO type of things uh there is you know standard kind of uh call to actions adaa Frameworks copyrighting Frameworks things like that but uh yeah there's really no reason you have to do this inside of the document editor other than you end up with a document over here on the right hand sidebar like I just had a create some lyrics for me about lifetime deals at appsumo and it says a world of fleeting Trends where the value often bendz there's a place where the deals
• 17:00 - 17:30 are real and lifetime offers seal the deal so we've got a full song here written for us abmo where dreams come true lifetime deals just for you grab the chance don't delay Make Your Mark start today okay so that's a document editor like I said the least exciting tool that we're going to come across for the rest of this video well with the exception of maybe the image generation which kind of sucks uh the next section however is really cool we've got all of this uh PDF engagement stuff and then there's also o chat with YouTube so
• 17:30 - 18:00 let's go from left to right we'll just go upper left hand corner create flashcards you can take a PDF upload it and it will generate flash card cards for you so let's try this out I'm going to click generate right here and then I will upload either a YouTube video a PDF or a website I happen to have a PDF ready so I'm going to choose that all right so the PDF is called math fundamentals I'll click next select the number of flashcards I don't know let's just do 10 for now I'll generate the flashcards and here here we go I've got a set of flash cards now unfortunately
• 18:00 - 18:30 the first one is kind of about the people who put together the document so we'll skip over to flash card number two and then I've got some questions about mathematics so I'm just previewing the flash cards right now so I can see both the question as well as the answer but if I wanted to start studying I click this button and I will only see the question until I'm ready for the answer so like in this scenario it's giving me multiple choice I just saw the answer so I can just click that and I'm ready to go so for some context here I literally just downloaded a random PDF on the
• 18:30 - 19:00 internet called math fundamentals and like the whole first section here is about the company that put together this PDF it it's very long it eventually does get into some actual math problems here but the flash cards that I seem to have generated are mostly based on the first part of the PDF so that's kind of a bummer there is a little friendly assistant here called Orion it's Orion the koala and it's a flashcard assistant where you can ask questions about the flashcards so assuming you had good
• 19:00 - 19:30 training material I think your results are going to be a lot better here I probably wouldn't upload like a 700 page PDF it's just too much for 10 flashcards I mean if I'm going to be critical I'd say that like hopefully the AI will be smart enough in the future to know that I don't care about the context of where the PDF was generated from or who made it I want to actually study the information so in the end that's kind of more on the llm than it is on the tool here that I'm using but uh yeah I I would say definitely strip out any
• 19:30 - 20:00 information you don't want to be used to generate the flashcards before you upload your documents okay let's move along to chat with YouTube this is a really cool concept you're watching a long video you don't have time to finish it well just let me have a i summarize it what's the range of Revenue all this does lar the blue color billionaire I'll pop back over to chat playground drop in the YouTube url load up the video it's going to take a second here while it transcribes the video
• 20:00 - 20:30 I mean it was literally a second it was not long at all and now I can ask a questions about the video so I'm going to ask the most obvious question the thumbnail says I made 600 million a year doing this well what is this what was he doing let's go ahead and ask says Larry built one of the largest and most successful basement repair companies in America and has a portfolio of blue collar businesses that generate $650 million a year very cool I'm going to say what do you think was the most important important takeaway from this video for a regular entrepreneur and the
• 20:30 - 21:00 AI lists three things first of all was having a positive mindset then it says uh building a great team and then the notion of incremental process rather than seeking quick success all right those are all very helpful things I think I'm good with the video I don't need to spend the 12 minutes sorry no you guys should definitely go watch this though all right next up is chat with a PDF this is exactly what it sounds like you upload a PDF and you can ask questions about it oh I try to upload the math fundamentals PDF and I got this
• 21:00 - 21:30 err here so not good now to be fair this is like a 27 megabyte PDF so it's kind of chunky that's kind of long let me find a shorter PDF all right I'm going to upload this Old Dan Kennedy PDF here and ask some questions about it all right it is uploading and this one worked just fine I really think it was the size of that PDF that it just couldn't quite handle it all right so I'm just going to ask some simple questions here like it says that this is published six times a year every other month let's find out if
• 21:30 - 22:00 it knows how often the document is published this is a newsletter how often is this published it's published every other month perfect next I'm going to ask it about how many times people are dining out per week There's a little clip right here it says Uh there's a trend of people who were dining out multiple times per week dropping down to only once a week and then it predicts that this will get worse and people dine out less so the idea here is obviously like you remember a little bit about article or about whatever you read but
• 22:00 - 22:30 you don't remember exactly what it was and you don't want to go searching for a specific word inside of the document so instead the AI can read it all and you can just ask questions about it so here we go what does it say about how often people are dining out document says that there is a trend of people who were previously dining out multiple times per week dropping down to only once a week it also predicts a furtherance of this trend so nailed it so that is chat with PDF the last option over here is called AI notes and it can generate concise and informative notes from PDFs let's use
• 22:30 - 23:00 that exact same newsletter and just have it take some notes for me I could also do this from a YouTube url by the way all right I've uploaded the PDF let's generate notes and boom there we go we've got our notes sure you can do this over in the playground by just copying and pasting everything but this is pretty easy too all right two more features to get through one is image generation I'm not going to spend a lot of time here here is the image I had to generate earlier of a sumo wrestler eating a taco it it does feel stable diffusion to me it's not terrible right but it's not up to
• 23:00 - 23:30 the standards that we're seeing from some other tools right now all right so I'm going to say generate an image of a shrimp with a cowboy hat and the style I'm going to use is let's do illustration and generate image says images can take up to 30 seconds to generate and there we go actually that one is pretty good I may have written this off a little bit too soon if you do need some shrimp with cowboy hats this is your tool the last feature of chat playground is called mixture Ai and this
• 23:30 - 24:00 is a very interesting idea it's going to take chat gbt Gemini Cloud as well as llama and combine them all to give you a mixed output so it's essentially going to analyze what the best parts of each answer is and then give you a final conglomerate of all of them in a single output so rather than just picking and choosing between the different ones it will combine them for you all right so I'm going to say write me a message to my client about how grateful I am for 10
• 24:00 - 24:30 years of their business I gave some little details about the business here let's go ahead and see what it does grabbing information from all of the major llms all right so you can see each llm working here and now once they are complete mixture is going to take over and combine them all into a final output which should be the highest quality version all right so I've got my mixture of output and I got to say I'm not super impressed because it starts with in my
• 24:30 - 25:00 opinion the biggest fatal flaw an llm can make I hope this message finds you well that there's nothing more obvious from an llm than that line notably the only model that chose to start that way was Chachi BC I hope this message finds you well so unfortunately it picked that one to go with everyone else started beautifully by the way Gemini starts with it's hard to believe it's been a decade since we first partnered with Joe's widgets uh Claude says I wanted to
• 25:00 - 25:30 take a moment to express my heartfelt gratitude for your incredible loyalty and partnership over the last 10 years and llama says I wanted to take a moment to express my heartfelt gratitude for the opportunity to work for you over the past decade so interestingly that llama and Claude are pretty similar in their output here but I'd say all three of these were better than Chach BT which means it's also going to be better than mixture just because of that one line now I can obviously edit that out but uh yeah I don't know I'm I'm not sure I'm
• 25:30 - 26:00 sold for this specific purpose on mixture yet I'd have to do more testing to really know for sure where I think it will probably improve is in the form of doing like problem solving coding anything that takes more logic having more inputs and then kind of weeding through the answers I would imagine that mixture is going to do a better job in that scenario okay so that is the entire tool let's go through the plans and pricing and then I'll give you my final score so everything you've seen so far in this video is performed with tier one the $59 plan that's a onetime fee you
• 26:00 - 26:30 get access to this tool forever and you get 300 queries per month it's only for one user if you go up to tier 2 all of your limits essentially double but you also get the benefit of being able to bring in your own API key so that way you can continue using the models just pay whatever they're charging and you won't ever have to be worrying about going over your limits and getting shut down of course there's going to be extra fees but those fees are going to be relatively small and usage based now if
• 26:30 - 27:00 your own AI key is not enough for you you want to have everything bundled into one payment there are other plans or maybe you have a bigger team than just two members and you need more seats you can go up to tier three get five seats and more than double the amount of queries but only more than double by 50 there is also tier four which gives you 2 200 queries per month and 10 seats and then finally code five gives you 5,000 queries per month and 25 team seats so a pretty beefy plan right there all right
• 27:00 - 27:30 time for a score chat playground AI is going to get an 8.1 out of me I think this is a very impressive tool there were a few bugs along the way but a lot of times that's going to be because of the API integration sometimes things just go down or they have limits and those things will be hit occasionally but I do think this is one of the best implementations of this type of tool that I've come across not just in the LTD space but really anywhere this is an exciting tool and I could see it even replacing our expensive Claude teams
• 27:30 - 28:00 plan we pay 150 bucks a month for that I could get a one-time purchase here of I think I'd be able to get by and maybe tier three gave me five seats that's all we need what would that be 239 bucks that's less than 2 months of colla teams and I'd have a lot of different tools available to the team so clearly a lot of value here yes we do have to worry about longevity with a tool like this because how are they going to pay their own API bills but like I just mentioned 2 months and I'm already kind of on the
• 28:00 - 28:30 profitable side of things in the instance where I actually do need five seats if you are worried about your longevity with your LTD purchases this is a reoccurring theme I'm seeing it more and more in the comments I want to make sure everyone knows about the appsumo buyer protection plan what this means is if a deal goes down within the first 6 months meaning that they go out of business you will get 50% of your credits back if you're not a plus member and you will get 100 % of the credits back so an abso store credit not your
• 28:30 - 29:00 money back you will get a store credit so you can buy something else to replace the failed deal 100% of that money if you're an abso plus member and the deal is an abso select deal you can always tell if a deal is a select deal because it says select right here in fact what I do when I browse abso is I just go to software and then I choose absolo select I don't even look at anything else that's the way I do things you do you I'm not trying to be negative here I hope chat playground sticks around for a
• 29:00 - 29:30 very long time I think they got a great tool here I don't see any reason they couldn't succeed outside of the LTD space but it is good to have the purchase protection policy in place in case you actually need it all right let's wrap this thing up I'm Dave Swift once again from client amp.com if you want to check out the tool we've been talking about there is a link Down Below in the description or you can support the channel just by clicking on that anytime you make a purchase over at appsumo head over to client am.com get signed up for the free email newsletter leave me a question down below like And
• 29:30 - 30:00 subscribe all that good stuff until next time I'll see you in the next review