AI powered Super App is the future

Añadido: 2026-05-15

[00:00:00]If I ask you what is the most used app in the world, chances are that you'll say Instagram, TikTok, maybe even YouTube. But the application which is most used in the world is WeChat. It has over a billion users doing everything that you can possibly imagine from booking tickets, cabs, making payments, doing everything that you do in a day-to-day basis. That's a super app.

[00:00:21]For a big part of history, a lot of Western applications and companies have tried to create a super app for the West and generally for the world as well. And we do have some applications, fragments here and there which promise to do everything, but I do feel that AI brings certain amount of confidence in that promise and in that vision. And in this video, I want to discuss a little bit about the super app which can exist with AI and what does the future look like?

[00:00:46]So, see, in order to understand this, something I've been thinking for a long time, let's talk about these super apps, right? So, we have, for example, this WeChat in China, mostly, which is the super app, right? So, this has been like something that people there use, over a billion people use for everything. But at the same time, if you talk about West, or basically any country, you will have different app for e-commerce, different for payments, different for search engines, for example, taxi booking are different apps, movie tickets are also different, and so on and so forth, right? Now, people have tried to consolidate and combine this, but all the companies and all the startups have failed. What I'm seeing now is a category of applications which is possible with the help of AI, right?

[00:01:30]So, AI unlocks an interesting use case which we don't really talk about. See, first of all, I feel that AI is conversational, right?

[00:01:39]And I don't feel that this is the end UI, that is just the chat. So, text is obviously not the end UI, but conversational UI is actually great because it unlocks that human element of just asking somebody to do some work for you. And I don't know if you know, but a lot of this work, you know, you buying groceries, for example, or you making payments, or you know, you just doing your flights, or you know, maybe tickets, or airplane tickets, and so on and so forth. If you are sort of like ultra-rich, which, you know, some of the people are in the world, they usually don't do any of this by themselves, right? They have their personal assistants, uh they have people around them who are taking care of everything in their life, in their you know, they just get handed over a schedule that you have to be this time at this place, and they will just go there, they will earn their money, whatever, from the sponsorships, from the deals that they are doing, and rest of the stuff is taken care of other people over here.

[00:02:33]Now, I'm not saying that AI of today can do that, but that is an angle at which I'm thinking, you know, a sort of super app, which is still conversational with you, which analyzes and understands your whole life, and is able to provide you with interesting insights and stuff, things to do, for example, and you can in return also ask it to do things, for example, booking flights, hotels, etc. You know, doing a bunch of things, connecting these connectors, adapters, and so on. In a way, if you think about it, applications like WeChat are popular because they are complex and have everything, right? AI-based apps are generally much simpler because the complexity is hidden behind a conversational UI, somebody you can just ask everything, and they can possibly understand it, do that work, or let you know why that is not possible.

[00:03:23]For example, very simple example of OpenClaw, right? One of the AI agents which has gotten a lot of hype. This is sort of an agent or a system which could be part of one of the super apps that you have, right? Or maybe like, let's say something you need for editing documents, or you know, building, let's say, presentations. So, if what if you can ask your AI agent to do that for you? Or even coding, as a matter of fact, right? If you're building like applications for fun or for business, that can also be part of just one super app, right? Right now we have a bunch of different apps. For example, for coding you'll have VS Code, for docs you'll have to pull it into Google Docs or, you know, Word, MS Word. For presentation then you have different tools and so on and so forth, right? So, what I feel and what I see is that AI has that potential, when done right and when done properly, it can not only combine all the existing apps, the most common ones at least that we see, into a single application flow, but it can also reduce your need to have multiple applications in the first place, right? So, it's it's this super app is not like a combination of a bunch of apps. It's actually a reduced combination of complexity while increasing the productivity that you have on every individual app. And that is where today's sponsor, Chat LLM, also comes into play from Abacus AI. Now, this is a product when they reached out to me, I took a look and it figured out that this fits very nicely into this topic of discussion that I wanted to talk about, which is a super app, because they aspire to create exactly that. And everything that is possible with LLMs and the world of AI today, you would be able to find on Chat LLM. It's built specifically for teams, but you can also start as an individual with just $10 a month. And even for your first month, if you'll use my link below, you'll get $7 as the starting price. Now, this is a product that gives you, first of all, multiple models. As we know, some of them are cheap and fast like Deep Seek. Some of them are expensive and, you know, state-of-the-art like Opus 4.7. And maybe like, you know, image models like GPT Image 2, for example, which is a great model for generating images. So, all of them are here. And this route LLM that is there, which I'll show you, this is sort of the nerdy stuff, right? So, this is sort of the super agent sort of thing that we can talk about because this can route to any large language model which it feels deserves this query. For example, let's say you are having a conversation, this can route it to Claude on it 4.6 because the conversation is not very deep as a matter of fact. But then suddenly you want it to generate music for you. I don't know. Or, you know, images or video for you. Then it can route to OpenAI's latest models for doing that work exactly. And because this is a system that supports like the whole agentic flow as well, you can take your Open Claw instance that I talked about and deploy it on Abacus AI, which they call as a Abacus Claw, which basically gives you all the benefits of Open Claw, but you don't have to deploy it on your own. There are a lot of things inside Abacus AI itself, which makes this a perfect candidate for a super app of today, which you can use for image models, video models, even Open Claw having a core work sort of experience with projects, doing code generation, wipe coding websites, everything. And as you see, once you log in, you would be landed into an app store of common applications that you might want to get started with. For example, LinkedIn outreach or building a website with payments integrated or, you know, doing your AI resume analysis. Or maybe even learning a new topic, right? If you're learning something new, you might want to have AI teach you that. So, I have a prompt that I want to check over here, which is basically, can you create an Instagram cover, which is I should probably specify 9:16 for a video topic, which is would there be a super app like WeChat but built as AI native core. Less text, more graphic is preferred. And I'm going to select route LLM over here so that I can see it in action how it can route it to a specific LLM, which is image optimized. Once you're done, click on send and wait for it to work. And as you can see, it has routed it to GPT 5.4, which has in turn routed it back to GPT image two, which is interesting because GPT image two is indeed the best model today to generate images and which is exactly what I would prefer. So, what happened here is that GPT 5.4 is understanding what needs to be done. So, it would create the final prompt in the first place, but the image generation model would be GPT image two, which is the one which actually is good at generated generating multimedia. So, that is where the super app sort of also starts to come into play, right? For example, you have an application for booking tickets and then there is a large language model. I'm just making it this up, but let's assume that there is an LLM which is extremely good at browsing websites. So, it knows like, you know, I don't have to click on malicious links or I have to browse it in a specific way. Maybe I can also use networks tab to inspect HTTP calls or whatever. Right? It's it's like a really good web browser LLM. And at the same time, there are other LLMs that are able to call it, right? So, that is where a sort of super app consists of logic consists of as well, where there are multiple AI agents can be of different LLMs which are orchestrating with each other. So, as you can see, it has created an image for me which has these logos of, you know, things to do which is sort of like a super app concept.

[00:08:34]Now, I'm obviously a little bad at prompting in terms of specificity, but you can specify exactly what you want.

[00:08:41]Maybe even make some changes and it can do it for you. And of course, the best part about this is that if any of these providers is actually down, then this would automatically route it to an LLM which is actually up and running. Which happens more often than not with at least cloud models that we have seen so far. And mind you that all of these models that you see over here, even the open source one like Kimmy and Deep Seek are hosted the open source ones are hosted by Abacus AI themselves on US servers. So, your data actually does not go to China or any other place for training purposes. These large language models are perfectly fine to be used inside your teams with sensitive data like your code, for example, or like like your internal ad campaigns or whatever you want to do. And this would be routed to Abacus AI only. Let's try another prompt. Can you create code for FizzBuzz program in C++, right? And I'm still going to use route LLM. And for this, it has routed it to Claude 3 Sonnet 4.6, right?

[00:09:38]Which is fair because Claude 3 Sonnet is smart enough for a simple problem like this, and it's also not cost hungry, right? So, it will not cost you a lot for this program, which is, by the way, it looks fine. Of course, if you're not happy, you can also regenerate it with Opus 4.7 or any of the expensive models.

[00:09:54]So, this is sort of like a super app for working with all the models, all the things that you might need like image generation, video generation, even speech, right? So, you can have text-to-speech where you are using state-of-the-art models from Eleven Labs, OpenAI, and Hume, and you can specify whatever you want. So, it's like an all-in-one solution for you as well as your team, which is, by the way, super easy to configure. All you have to do is go to your profile over here and go to the team section and add people to your account. And look at this, they are also introducing new verticals, which is like health, legal, role-play as well, which just touches upon the fact that there could be a huge market for super apps which include AI as a native code, right? We might see applications like Chat LLM and built by Abacus AI, which are applications that you can use to do everything. And whatever you can technically do today is possible with Abacus AI. So, do check out Abacus AI.

[00:10:49]All the links are in the description to get started with it. I feel that this is a great add-on if you are juggling and struggling between multiple applications today. But yeah, that's what I feel about the future of super apps in general, right? We don't know like what exactly will happen because everything just changes every 6 months and so. But if you want to stay updated and if you want to be informed of how these things are, I feel that you have to stay connected with the broader ecosystem and with the changes that are happening through content like what you're watching right now, through actually using applications by yourself as well to see how good these models are getting. And if you subscribe to a super app like Chat LLM by Abacus AI, you get to use everything across every modality, whether that's audio, video, image, and experience like how this thing can be constructed at some capacity using AI.

[00:11:39]Of course, web and automating web is a big part to still solve because there are platforms like Cloudflare, which are literally getting paid for just blocking the bots. And in the past, bots were just malicious actors, right? Now, the challenge is new because bots can also be actually useful AI agents. So, there is a system that everyone is developing around payments, micro payments, you know, how we can authorize AI bots to actually validate themselves and, you know, tell them that they are trusted AI bots, tell the website provider. So, there's a lot to still figure out in the infrastructure world of browsing the web at least. But from an intelligence point of view, I think we are there where AIs would be able to take away a lot of your personal hygiene work from you over a simple conversational UI. So, that's all for this video. Hopefully, you liked it.

[00:12:30]If you did, make sure you leave a like and subscribe to the channel. Do check out Chat LLM by Abacus AI, which is as good of a super app that can exist in today's time. And you can get started just for $10 a month. You'll find all the links in the description. That's all for this video, and I'm going to see you in the next one very soon. If you're still watching, make sure you leave a comment I watched till the end below to tell me that you were still here. And let me know what do you think about the video.

[00:12:58]>> [music]