How to Build Full-Stack Apps With Verdent AI | Parallel Agent Coding

So guys, what you're seeing on the screen right now is a complete, fully functional personal finance dashboard with budget tracking, transaction categorization, spending charts, the whole thing. And I built all of this from scratch in one sitting. That's without switching between a bunch of different tools, without it being built on a boilerplate template, and with multiple parallel AI agents used to build it. So in this video, I'm going to show you exactly how I built this app. We're going to build it in real time, step-by-step, and then I'm going to show you how you can do the same to bring your ideas to life with AI agents. So the platform that we're going to use today is Verdant. Verdant is an AI native coding platform that's built around the idea of multi-agent orchestration. So instead of giving you one AI agent, what it does is it gives a manager agent the responsibility of delineating tasks to specialized agents that can work in parallel or together on different parts of your code base. So instead of, let's say, with FOD, right, one agent building one thing, you have many specialized agents working in parallel on your idea or project. So first, as far as what we're solving kind of with this update or what I want to touch on, and you'll see this in a minute when I talk through the platform that we're using, but essentially nearly all AI coding platforms today, they are one agent that does one thing at one time. And that can be a limitation because in theory, why should there just be one? Why can't there be three or five if that one is really good at its job? And so that in mind, the platform that we're using today is Verdant. And it was founded by the former head of algorithms at TikTok and the former head of tech and product at Daidu. So it's a pretty cool team building this thing. So we're going to jump into my computer to get started. Now, if you want to follow along and build an app with me in this video in real time, you can do that completely for free. There's a free trial on Verdant. That's no risk. You can click the link in the description and you help support the channel as well. If you don't want to build along in real time, no worries at all. And you can just keep watching that in mind. Let's go. Okay. So here we are in the Verdant desktop app. What I want to start by doing before we get into the first component of our build is just set up the project. What we're going to do right here is select new project. Then I'll click new folder and I'll call it personal finance dashboard. And this will be hosted directly inside your desktop. So now we're working out of that folder in my computer. And then I'm going to change under agent to plan. And this is component one. We're not going to get directly into the build. We're going to let the agent think through and actually plan holistically the app or the dashboard that we're going to build. And then we'll give it this prompt. So I want to build a personal finance dashboard web app and it should have these things. And then I'm clean modern UI with dark mode and I want it to be developed or coded in these languages. We have ultra think turned on. That just means it'll plan really, really well. Let's go ahead and click start. And it's going to start by asking us a few questions. What I'm going to say is no backend. We'll just do local storage and then we'll do chart.js. I'll click submit. Okay. And here we can see the result of plan mode. We have a full plan that we can scroll through and we can see here there's a lot to go through, but basically we can see the objective. We can see a visual architecture overview. So we have an app, dashboard, budget tracker, transaction forms, spending charts, and we can see the more detailed structure. So it's thinking through the format of the project, implementation steps, basically the guidebook on how it can go and actually build this before it starts to do so. So if we want to change anything, we can do that right here by just engaging with the agent. Otherwise, this is the plan that the future agents will use to better build the actual app, because it's just following a plan instead of trying to think and build at the same time. So now we can actually start our build following this plan. And this is where the parallel agents start to kick in. So what I'm going to do, I'm going to just click build right here. So in this chat, it's going to start working on it. And now this agent has started the work on the project. Now what I'm going to do while this agent is running, I'm going to click new task right here. So we're the same project, but I'm going to say, think about new features we could implement beyond this build plan. So same interface, very minimal disruption, but we can see we have one agent still running here, still coding, and another that's being started. So that's two agents that are being run. What if I want to start a third? And what I'll say for this one, and let's change from Opus. Let's go to a cheaper model. We'll go 5.3 codecs. And I'll say, while the other agent is building the app, go ahead and review the code base in real time for bugs. That brings us up to three parallel agents that are working for us. So here are some of the new features it was thinking of. And we have the here, and then it's still coding right here, and we can see those projects in the top right. So now let's take one of these features. Let's say number three, data import export. And then I will open a new task. I'll say, work on adding this feature. We're going to start that. Let's jump back into future feature brainstorm and say, think of more. And let's go into plan mode, and I'll say, build out a plan, let's say for your favorite potential new feature. So now we have four agents with four frontier models simultaneously working on building our app for us. What that means, fewer hallucinations, fewer errors, and faster as well. Now something else I think is really cool. I'm going to start a fifth agent here, and I want you to note this skills tab right here. And what a skill is, is a very specific material capability that you can add on top of the base agent that we are running within our project. So I'll click skills. We can see a lot here, just a CLI that lets the agent operate in your browser, CI helper, a lot of Figma stuff, which is awesome. What I want to do is install the web app testing. So I'm going to go ahead and click install, and let's go install only. And now that skill is installed. So I'll say, use your skill to test this web app, and let's just see what it says and how it incorporates that skill. Also, if we jump back over here, it found the savings goal of the most interesting. It also made another plan there for us. Let's go ahead and just click build, and now it will work simultaneously. So we have five agents simultaneously building for us. And we can see over here, it loaded the web app testing skill, and then it's going to run targeted checks to see how that initial build is going. One note on the UI as well, if we go up here, for example, and I just select this text, I can either add that to the chat, or I can create a new task in a current workspace. So still in this workspace, just based on this. I can even do that if I go into the code. Let's jump into this file. So I've opened up this file. I'm going to open all of this right here. And then I can say, go ahead and add that to chat. So I'm referencing these specific lines in the code. And then I can say, for example, troubleshoot this, or maybe I want to change something specific, like don't use read-only there. Whatever change I want to make, all I can do, or all I have to do to make that change, is I go, I select the lines, and I directly add that to the chat. We can also open as many tabs as we want here. So if I jump under Acorn, for example, see, I can open that, and there we go. And you can see, one of these files is 884 lines, and we have a hundred, a few hundred files here. So there's a lot of work being done. And remember, all of this is being done in the context of our computer, in that project that we set up. So if I go to that personal finance dashboard, then I can go ahead, and I can see those files actually being added there in real time. Okay, and here we can see that the personal finance dashboard is now complete. And this is where we get to the third big component of the demo. So number one, we planned. Number two, we built. And number three, we're going to review. We're going to let multiple agents in parallel, comprehensively look for bugs, and make sure the app is ready to go. And in the same style as the build incorporating parallel agents, this will reduce the probability that bugs get through, and produce better apps like so. So what I'm going to do, I can go to the local host to check out the app. But first, I'm going to click under here on multi-model review. So I'm going to press on that, and this is going to have multiple models have kind of a roundtable discussion. So they'll each review the app and the code base. They'll talk to each other to kind of compare notes. And then once they decide, and they settle upon bugs, or improvements, or how to proceed and fix those, then they'll go through and make those changes, or just make them available to me. And a quick shout out here, I mentioned the next action feature. What this is doing right here is saying, hey, we think that what you're going to want to ask next is to play this to production, or add a search or filter feature for transactions. So it's proactively saying, hey, do you want to do this? And if we do, we just click right here, and it'll do that automatically for us. And that applies across all of the messages that we can see here in agents. For example, if we jump into this chat, which is no longer generating, I can say, test the savings feature, or run the dev server. For example, let's just let this run as well. Let's say, go ahead and test that savings goal feature. And our models are still reviewing over here. So it finished the review. Now let's jump into local hosts to check out the final build. And so here is our final build for our personal finance dashboard developed with multiple agents, planned, and finally reviewed, also by a round table of agents. All of these work. These are fully functional savings, goal, budget, transactions, pages. We can see the import from CSV or Excel feature that we added right here. We can also manually add a transaction. And these are fully functional features. This is not like a demo. Let's say I want to add $1,000 of income for my, let's go dog makeover business. We'll say that's tomorrow. And here we go. Here's my expenses, or excuse me, my income for the month. Now we'll say we spent $600 on rent, description, just rent. I'll add the transaction. And now we can start to see that data is getting filled in. That syncs across all the pages and we exceeded our monthly budgets for the month. But if we want, we can go ahead and change those budgets. So this is something you could actually use. You could customize to your life. And we built this in about 12 minutes using all of these frontier models and there are no errors, no bugs. This just worked. So let's finish this up by creating a goal. And that goal will be you go build something now on Verdant. That's my goal for you. After watching this video, that will be a $10,000 goal for right now. That's going to be other custom color. Ooh, we have a whole color panel. Actually, that is pretty epic. We will create the goal and there we go. We have created a goal and now it's up to you to go execute on that goal. So there you have it. That is the final product and all the features inside Verdant that I wanted to show you. My overall take here is the future is lots of specialized agents working together because if one agent is really good, why shouldn't multiple be even better? And so Verdant is on the cutting edge of using agents together, whether in committees, whether in parallel processing to help you become more productive as a developer, as a professional and as an entrepreneur. So again, if you want to support the channel and try out anything and everything you saw in this video completely for free, you can click the link in the description to try the free trial of Verdant. That in mind, I hope you all enjoyed. I appreciate you spending some time with me today and I'll see you in the next one. Cheers.