Transcription
Wow, this looks good. I'm pretty impressed. Chat LLM is one of the cheapest ways to get pro-level access to all of the state-of-the-art models and features, including tasks, projects, and automation systems. ChatGpt, Claude, Perplexity, Nano Banana, Midjourney, and so many more. You can test out all of the AI models in one place to figure out what works best for you. The route LLM feature even automatically analyzes your prompts and suggests the best model to solve your specific problem, so you don't even have to think about it.
Hi, my name is Callum, also known as Waterloots, and welcome to today's video on Chat LLM by Abacus AI, the all-in-one AI tool. With new AI tools and advancements happening every day, it can be incredibly easy to get lost in the noise of all of these different tools and features. It's hard to know what the best tool is for your particular use case. Chat LLM helps you cut through the noise and access the tools you actually need to save time and money while boosting your productivity. Abacus AI actually reached out because they align so strongly with the message that I've been teaching on this channel on how you can use different AI tools and help people cut through the noise and make AI more accessible for everyone. If you're interested in checking it out, I have a link in the description and the top-level comment so you can either follow along or learn more. It can be found at chatlm.abacus.ai/cwl.
In today's video, I'm going to walk through an overview of Chat LLM, including all of its benefits and features. I'll walk through example use cases, including for research, writing, presentations, and image, video, and code generation. Then I'll get into projects and custom AI tools, before touching on tasks and automation, and finally sharing some tips for conserving your credits and making the most out of this incredible tool. A reminder to please like, hype, and subscribe if you find this video helpful. I really appreciate your support as you enable me to continue testing out these tools and making these videos that can hopefully help you as much as possible. So, thank you. Now, let's dive into Chat LLM by Epicus AI.
So, Chat LLM is actually pretty great because it's a single AI assistant that gives you access to all of the different state-of-the-art LLMs. They've got Chat GPT, Claude, Sonnet, Gemini 2.5 Pro, Grok 4, and a whole bunch of other ones, which I'll show you in a second. This is Chat LLM Teams, which means that not only can you use this for yourself, but you can also connect different users into the same user base so that anyone can use it. So for $10 a month, you get access to a lot of the features that only come with plus versions on the other LLM on the other AI platforms. So with Chat LLM, for example, you get access to GPT-5 and it has access to reasoning. In order to get access to that here, you need to pay $20 a month, which is already double what Chat LLM costs. This also gives you access to projects and tasks, which also is available. Deep research and agent mode. Again, these are all available in Chat LLM. That would cost you twice as much with Chat GPT. Perplexity Pro costs $20 a month. That again costs double what it costs you to use with Chat LLM. Google AI Pro, it costs you $27 Canadian a month. Claude Pro also costs $20 a month. Nano Banana is $15 a month to get access to image generation. Runway is $12 a month. So the idea here is that you get access to all of these different tools, but you don't have to pay $20, $40, $60, $80, $95, $107 to get access to it, plus more products. Instead, you just pay the one $10 fee and you're good to go. And you get 20,000 credits a month with Chat LLM through Abacus AI membership.
One of my favorite parts of Chat LLM is that you have full control over your data. They don't use your data for training their model or other models. Something to keep in mind as you use any AI tool is what are they using with your data? And it makes me more comfortable to use a tool that I know that I have full control over my data. Let's get started.
So, when you first sign into Chat LLM, this is what your homepage looks like. And this looks pretty similar to most other chatbots like Chat GPT, Perplexity, any of the major ones. There's a whole list of tools here we can get into. There's apps and tasks and deep agent. We have projects and chats on the side here. So, my thought is I'm going to give you a quick overview of all of the various features and then we'll dive into some more specific use cases.
So, the major benefit that I've come across with Chat LLM by Abacus AI is this route LLM here. So, if we click on it at the top, the idea here is that with this single subscription that's $10 a month, you're able to get access to all of the state-of-the-art models and use the route LLM to automatically select the most useful one for whatever your particular use case is. So, you can kind of think of route LLM like a compass. If I go and I want to ask a question like "How will knowledge management change with the introduction of AI agents?", I can click go and you can see here that route LLM routed to GPT-5, which is OpenAI's latest model. So what's cool here is that you don't have to think about what model do I need to use to get the best result here because route LLM is going to go through and select the best model based on your particular question. Let me just switch this to dark mode for a moment. There, that's better. And I can just keep going and having a conversation like I would normally. And you can see down in the bottom right here how many credits are used. Different models will give different credit usages. For example, maybe GPT-5 takes more credits than GPT-5 Mini. And honestly, like GPT-5 just did a great job here. But if I wasn't quite happy with it, maybe I want to see, oh, you know what? I'm going to regenerate this using Claude Sonnet 4. What this does is it lets me compare how the different models are used. And this is why I'm saying if we go back to the homepage here, you can kind of think of route LLM like a compass where depending on what your question is, it's going to point you to the best model for your particular use case. So you don't have to worry about clicking through all the different models trying to figure out if you should be using Gemini, trying to figure out if you should be using Claude or OpenAI. Instead, you get all of the models in one place and you don't really have to think about it. So I'll show you more of that in a moment.
Let's take a look at a couple of the other features. So we can see on the left-hand side here that we have the chat, which is just a history here. For example, here's my question I just asked on AI's impact on knowledge management. And we also have this option for projects here. So what's cool about projects, let's say I just go into my generic research one that I just created. I can go through and I can introduce custom instructions that automatically will apply to every single chat I ask in this project. Similarly, I can upload files for context in the project, which kind of turns Chat LLM into its own knowledge management system because I can add project files and then ask questions about those documents, about all the files that I'm uploading. So again, I'll show you more on projects in a moment. We also have then these options down in the bottom left here for deep agent and apps, which is where you can actually build and deploy apps directly. And deep agent is what helps you do that. It's an agentic AI that helps you plan and organize code so that you can build applications and then launch them. Deep agent is something that would take a lot more to explain, so I'm just going to briefly touch on them today. If you're interested in learning more about what these tools can do, specifically Deep Agent, MCP, or Code LLM, please let me know in the comments because if enough people are interested, I'm happy to make a video that goes deeper into it. And then finally, we get into tasks. And tasks are basically automation where you can choose to run a particular prompt every single day, every week. You can control the scheduling and it will go through and generate a report for you if that's what you're looking for. It will send emails and for example, like I did with this one, scrape the internet for today's stock market news on tech stocks and then email me a report with what it discovers based on today's news. So that's a pretty cool feature and I'll show you that in a few minutes.
And then just getting back to the chat window for a second, we have the option here for all of these different types of prompts. And basically what these do is it modifies the way that you're chatting with this interface here to give you more customized outputs. For example, if I'm looking for an image, I can choose specifically which model I want to generate. Maybe I want to use NanoBanana. I want to generate one image. And I can give a prompt here. And I can just run and click generate. So you can see this is using NanoBanana. And this is where I didn't use route LLM directly to try and select the model for me automatically. Instead, I pre-selected that I wanted to use NanoBanana. So honestly, uh, a little creepy. I was thinking of this more as a robot. So, I can regenerate this image by changing to a different model. And this again is where the true power of Chat LLM comes in because you're able to modify and reselect and regenerate based on the output so that you don't have to keep switching between all the different websites. You can do all of this in one place. So maybe instead I want to use MidJourney. And we can see how MidJourney generates an image rather than how NanoBanana did. There we go. So that's interesting. It's showing just a different style of how this model, MidJourney, interpreted my prompt compared to how NanoBanana did. So I wanted to use this visualization here to show you visually how using different LLMs can significantly change the output that you're getting. And that's kind of the whole point of this system.
If we think of an abacus, which you might have seen these when you were a kid, the idea is that you can use different beads to represent different weights or different numbers so that you can calculate complex things. With Chat LLM, with Abacus AI, you're able to choose different models based on the task that you're looking for. You can choose different beads depending on whether you're generating images, code, if you want to build a presentation, do deep research. The idea is that you can use Chat LLM to automatically decide which model is going to help you solve your problem the best. And over time, what's cool too, let's say we go back to our initial prompt. I can regenerate this with Claude Sonnet. Just going to download a PDF of that for a sec so we have a direct comparison. I can see that that used five credits. But what would happen if we tried using Claude Sonnet? Let's try that again. So you can see that actually used eight credits. So it was a little bit more expensive to use Claude Sonnet than it was Chat GPT. You can go through and say, hm, did I like the numbered outline that Chat GPT gave me or did I like this more concise report that Claude Sonnet gave me. So it's completely up to you what works best. You can start to practice. You can decide over time what works best for you and you can start to assemble a team of different LLMs that you know you're going to use. So for example, let's say I went back and I wanted to create a new chat. Maybe I know now that I've gone through and found that GPT-5 did a better job for this generalized research. So maybe now if I go back to my chat, I can add a new chat and I can always specifically choose that I want this prompt to be run with GPT-5. I can turn on thinking mode. I can enable it to search the web. I can upload files if I need to. I can connect different apps. There's a whole bunch of information. I can connect Google Drive and bring that in. Really, there's just so many options here and I know that it can feel overwhelming, which is why I'm trying to show you roughly how this can be used at a high level before we start to get into some more practical examples.
So, let's say I wanted to generate now a PowerPoint. "Teach me about the impact of agentic AI and knowledge graphs." So, I've selected GPT-5. I selected thinking mode and PowerPoint. And let's see what happens. So, you can see it's given me the option to upload a PowerPoint template if I want a specific brand style, for example, but I'm just going to click the default for now and click submit. The planning mode there used up 33 credits. So, honestly, not that much considering everything that it's doing. Let's see how the PowerPoint goes. Since I've selected the default template, it's actually directing me to the deep agent, which does a better job with presentation design. So, let's click on that. And again, I'm just going to show you very quickly deep agent high-level for this particular PowerPoint feature. So you can see here that the deep agent now is going through and it's running its own independent searches. This is where agentic AI comes into play. It's not just a chatbot like an LLM. It's actually going out and running its own Google searches and then analyzing the search results to provide sourced material into the PowerPoint presentation. So that's pretty cool. Okay, so this honestly took the deep agent about 20 minutes or so to produce this PowerPoint, but I mean, I think it looks pretty good. It's gone through and generated all of these images and graphics to help teach people about agentic AI. And honestly, this is this looks good. I'm pretty impressed. So, I can click export presentation, export as PowerPoint, Google Slides, or as a PDF. But you can see here, too, uh, this used 3,133 credits. So, this one presentation did use up about 15% of my monthly allotment. But honestly, if you were trying to work on something for work and you needed to create a presentation on these topics, you'd be able to just drop in the files you need, give more context, ask it to update it, and maybe that would save you 10 or 20 or 30 hours worth of work. Like the detail in this presentation is honestly quite impressive. It even created graphs. Wow.
So, that deep agent that I just quickly showed you for the PowerPoint, that also works for deep research. And there's a whole bunch of different tools we can use here. For example, one we can use is called humanize. So let's say we go back here to AI's impact on knowledge management where we've regenerated using Claude 4. I'm going to turn on humanize. And this is where we have a few different options here where effectively what it's doing is it's taking the answers that the chatbot gave us and it's enabling different styles or instructions to apply to all the new messages in the conversation. I'm just going to leave this on automatic. Let's say, "Can you write a blog post explaining the impact of agentic AI on knowledge management?" So you can see again it's routing to GPT-5 and this was specifically using the humanize style. So the idea with this is that rather than just having some generic output, there's another filter being layered into this that hopefully makes it sound a little bit more natural, which I think can be helpful for learning and for explaining things to other people. And the idea is not to replace the human, but to make it easier. Maybe this will give me a great first draft for a blog post that I want to write. And here you can see at the end it's even suggesting, "Would you like me to generate a shorter LinkedIn-style version of this that hits the key insights so that I have some promotion material to promote the longer blog post?" So that's pretty cool that it's already automatically thinking about what should I do next with this blog that it's just written.
There's a whole bunch of different tools you can use here. You can use it to generate docs. We can get into scraping URLs from specific websites. We can analyze videos. Maybe why don't we try that for a second and go click new chat. Switch to the video analysis tool. I'm going to take my latest video, which is my year in review on YouTube. I can add this video URL here and click go. And now you can see because I chose route LLM again, it's choosing GPT-5, which is one of the most powerful recent models. So I'm not surprised it keeps using that. And now it's going through and it's analyzing the video based purely on the fact that I input the URL. So this could be a cool way to do research, to analyze long podcasts, maybe like Heman on neuroscience where it's a 4-hour podcast episode. Now you can use Chat LLM to analyze the video directly. Okay, there we go. So this just gave a great summary of my video that I just put out. And now I could, as a creator, use this to analyze what AI thought. You can see it's giving me strengths, gaps, nuances. It's telling me who the audience is: knowledge workers, creators, anyone with perfectionism or burnout, which is true. Again, there's just a whole bunch of information here. So, I could use this both as someone who is learning something new, again, effectively chatting with the video directly, or as a creator, I can use this to analyze my own creations and get feedback on it to hopefully make it better. And you can see that that used up a little bit more credits because probably it's doing video analysis.
And another cool tool that I've been playing around with more lately is video generation. And there are certain models here that you can use for free that I'll explain more at the end of the video. But for example, if I want to use V3. Wow. You can see that it says it might use up 60% of your credits. So it's good that there's that warning there. Let's try a different model. And just note that it might take a little while. Oh, here we go. I mean, I think that's pretty cool. Then you could take a few of these. You can, if you click generate video again, you can always upload or drag a file. So, I could drop in this file and then continue the scene where I left off. There's a lot you can do here.
As another example that I'm not going to get too deep into today, you see that there's a suggestion here for HTML landing page. So, if I click on that, you can see it's routing me to Claude Sonnet 4 because in my experience, Claude has been one of the best coding LLMs. And you can see it's actually going through and coding a landing page for me right here. It's just I'm not doing anything. It's just coding this whole thing for me. Great. And that just took maybe 20 seconds. And you can see here it's got a whole page. It's got all the buttons that can be connected to other features and we can just keep going and iterating. So from a web development perspective, this is also incredibly powerful because maybe you want to launch a business or maybe you want to make your existing website better. You can prototype and build directly inside of Chat LLM and then when you're ready, just download the code or there's an option to just directly connect to GitHub. But again, that's something that I'll get into more in another video if people are interested in the coding side of things. I just thought that was cool that you can generate the code so quickly and then preview it immediately. And this only used 73 credits.
Okay, so now that you understand a little bit more how the tools work, we also have the option to introduce projects. I just want to quickly mention that I am focusing specifically on Chat LLM today, but there are a lot of features in Abacus AI including deep agent and code LLM that enables no-code workflows and automation systems. So, if you're interested in me going deeper into those particular features, please let me know in the comments. I really appreciate any feedback. Also, if you're finding this video helpful, just a reminder to please like and subscribe and let's keep exploring Chat LLM.
If I were to go create a new project for a moment, you can see here it pops up and it gives me the option to give it a new name. Just call it "Wanderlude Tutorials." Space for brainstorming educational tutorials for my audience. And you can see here, this is the cool part. We get into project visibility. So I can choose: is this only available for me? Is this something that I want to share with the entire organization? Or is this something I want to share with other users in groups? So if I click "Share with entire organization," I then get more granular control. Do I want to share the conversations or just share the files and custom instructions? This is a really great way for you to begin connecting and sharing your knowledge with other people. You can have a team working in the same Chat LLM space that you give access to shared files and custom instructions on how chats should respond to you. So if route LLM is the compass that's helping point you in the right direction to use different tools, projects is like your workstation, your workbench where you're able to use and organize those tools for very specific use cases and then share them with other people. I'm just going to click this as "Only me" for now. Click create and we get this blank project space here. This is why I wanted to show you how the chat worked and how the tools worked first because you get access to all of those features within the project itself. So, every conversation I now have is going to start to get added to this list on the side, keeping track of conversations that specifically relate to Wanderlude's tutorials. And you'll notice that there's two other elements here.
The first is custom instructions. Basically, what this does is it takes your prompt, your instructions that you write in here, and it applies it to every single one of your conversations that you have. So, for example, I can say something like, "I'm an intellectual property lawyer and a patent agent with an engineering degree. I make YouTube videos about emerging technology, knowledge management, especially in Obsidian, and the latest AI tools to help creators, entrepreneurs, and knowledge workers in corporations or their own businesses." So, effectively, what I've just done here is these custom instructions have given context. So this context gets injected into every single conversation I'm having. I don't have to explain this context over and over and over again every time I'm working on a new tutorial. Instead, the AI will just remember this because it has the custom instructions built in. So this is one of the most powerful ways that you can get consistent output from an AI tool. So that even if we're using different models using the route LLM, every model will still get this overarching prompt that gets added to the conversation to try and keep things consistent because it has the context of who I am and what I'm doing.
We can also upload project files here to give more context. And when you do this, you have the option to add from files, which is the files that we've uploaded on the side here, or we have the option to upload from the computer. I'm just going to grab the conversation that we had earlier from Chat GPT. And this file here will now provide context for every single chat that I'm having in Chat LLM within the Wanderlude's tutorials project. Then just quickly to show you how this would impact the conversation, maybe I ask a conversation and tell it to search the web. "What are some of the best topics I can make a tutorial about that would resonate with my audience and provide them with the most value?" So, it should in theory understand the context that I gave it based on the fact that I added the custom instructions so that it knows I'm an IP lawyer and a patent agent and specifically that I'm using Obsidian, which you can see here. Now, I didn't have to go and explain to GPT-5 or to route LLM, which chose GPT-5, "Oh, I make videos on AI and Obsidian. I teach people about knowledge management." It just knows that already because I added that to the custom instructions. Cool. So, it just gave me a whole bunch of information which I'm not going to dive into because I think you get the gist of what's happening here. So, this is a cool way to just filter and streamline your knowledge management system a little bit more.
And with that in mind, let's take a look at tasks because that's another way that we can really streamline and improve the automation of these systems that we're building. So if chats are where you find your tools and use your tools, projects are where you build the tools and organize them, tasks is where you kind of set up the factory where you have this repeatable process that once you figured out the type of tools you want to use and how you want to use them, you can have them run automatically. So this is again another type of agentic AI where for example my instructions were "Analyze the stock market performance in the tech industry and create a comprehensive report." I have enable email alerts turned on. So, it was able to email my account directly. And then I have this scheduled to run daily. And I can choose how long I want this to run for if I want to or pause it if I don't want this to run anymore. Perhaps my credits are starting to run low. And I want to conserve the credits and not have my daily report. Maybe I want this every other day. So, I can choose daily, weekly, monthly, yearly, at a specific interval or custom. I can choose when I want it to run.
So, why don't we create a new task? When I click new task, I have the option to do deep agent or normal task. I could give it a task of "Can you do a weekly search for the most impactful advancements in AI and how it impacts knowledge management, automation, and efficiencies with high-quality outputs?" And you can see on the left-hand side that we've got this little task icon. So you don't have to do this through the tasks button. When you do it through the chat button, you just have to turn on the task tool down below. Let's click go and see what it does. Cool. Okay, so it's now said the task "Weekly AI Advancement Search" is scheduled to run every Sunday at 5:03 p.m. It will search for impactful AI advancements and analyze it. And it notes here too, "Running tasks frequently can quickly use up credits." So this is just something to keep in mind that if this is something you're going to be running every day or every hour, you might blow through your credits pretty quickly. And I'll give some tips on how to manage your credits towards the end. I can run the task immediately right now to make sure I verify the results. And I always have the option here to click edit. And I'm actually going to turn on email alerts. And I can choose what email I want to put in here. And then I can click view results. There we go. So we can see this used up 37 credits. It's just gone through and ran this search. And because I turned on email alerts, if I go over to my email, that might take a few minutes. So I'm just going to quickly show you this is the one that I got before on the stock market every morning. And it automatically produces the report, does all the searching for me, and then it emails me a report. So, this just shows you how you can start to use these tasks more effectively where I can schedule as many number of tasks as I want to as long as I'm just aware of how many credits I'm using up. And the idea is that if you find yourself doing anything repeatedly, you're doing something over and over and it's taking up any time or energy or effort and it could be automated, that means it's a pretty good idea for a task.
And what's cool too is if we go back to the chat, I could for example in my project within Wander's tutorials based on the different conversations I've been having, I could say, "What are some tasks that I could use in Chat LLM to make my process more efficient?" The goal here is not to use these tools in isolation with one another, but to understand how the different pieces fit together so that you can be more efficient, effective, you can conserve your time and you can conserve money. Because the fact that I'm able to use GPT-5 for searching the internet and Claude Sonnet for coding and Gemini 2.5 Pro for scripting and Grok for checking social media. You can use all of these tools all in one place and you don't have to worry about how much it costs for each of them. There's honestly no end to what you can do here.
And now there's just one more thing that I want to show you before we get into how the credits work a little more and some tips and tricks on how you can conserve your credits and make your $10 go further. And that's the option that if you want to, you can connect apps directly to your chat. So you can see there's Google Drive, One Drive, SharePoint, and Box. So this is a cool way that you can extend Chat LLM to all of your tools. I go to settings and click on connectors, and then there's an option to add any of these features to it. And you can even add MCP, model context protocol. So it just makes it so powerful that you can directly integrate your Slack and then you can start communicating with people and having people pass files through here. There are so many features that we could get into. It's hard to touch on all of them today, but if there are any features that you want me to explain more or explore more, please let me know in the comments.
Now, let's take a look at some tips and tricks on how you can conserve your credits in different situations. So, the big one that I want to show you here is that there are some models that you can use on an unlimited basis. So, this is GPT-5 Mini, GPT-4.1 Mini, Gemini Flash 2.0. There's a few here that you can use that don't use up any of your credits. So, as you're going through, depending on the level of detail you need or the level of complexity, if you realize that something's not going to take up that much effort, maybe you want to instead of just using the route LLM, select GPT-5 Mini as a good one, and that would then not use up any credits in your search. So, they have a list available in their FAQ under billing and you can take a look at these models and test them out and see which one works best for you.
Another thing to consider is that maybe in your tasks, if it's something that for example uses a thousand credits because it's a very complicated task, maybe you want to make sure that you're only scheduling that once a week instead of every day. You can just start to think about how you can use various tools within the limits of the credits that you get for the $10 a month. But for example, a lot of my searches were about 30 credits, which you can see that would give me about 666 chat queries back and forth with the most powerful LLMs on the planet. So that's pretty solid. Like I think for most people that's going to get them through the month no problem. If you're getting into coding, you might want to consider upgrading to the next level because that gives you more access to the deep agent. Coding is what's going to blow through your credits. Like if you remember coding this PowerPoint presentation used up 3,000 credits, so I only have so much that I can do per month with that. So the goal here is to be strategic with how you're using it. If you don't need to do a web search, don't do a web search. If you don't need the most powerful LLM possible, maybe you use a cheaper one so that you can conserve your credits for when you do need it. And I also think that it's great to just play around with it. Just test it out. See what works. I feel like there's so many opportunities where we don't know what an LLM is capable of until we try it. So the whole point here is to just experiment, get the data, and recalibrate. Just keep going, keep testing, and see if you can push the limits of what it can do so it can actually add real-world benefit to your life.
Chat LLM really is a great way to combine all of the best features of the different state-of-the-art AI models all in one place. I'm personally excited to continue exploring more, especially with Code LLM and Deep Agent, as I think agentic workflows are becoming more and more popular and more and more valuable for actual practical workflows. So, I can very easily see how this is going to fit into my system. If you found this video helpful, please like, hype, and subscribe. I really do appreciate your support as you enable me to continue making these videos. So, thank you very much. A reminder to please click on the link in the description if you want to explore Chat LLM more as clicking on that link does help me out. If you're interested in learning more how I manage my knowledge management system to be augmented by tools like Chat LLM, I recommend checking out my Obsidian and personal knowledge management playlist as I talk in many different ways on how you can organize your data to be augmented by AI. Thanks again for watching and I will see you in the next video.