How to Build With an AI Dev Team in 2026 | Verdent

Jon Law
434 followers
2 Video Views·Apr 8, 2026
Build With an AI Developer: https://www.verdent.ai/?ots=youtube&id=JonLaw
CLAIM 60% OFF CODE JOLA01
In this video: how to use the Verdent AI developer in both the dedicated workspace and VSCode version, whether for vibe coding or enhancing your manual coding.
CHAPTERS:
0:00 The Problem With Chatbot Coding
0:55 Introducing Verdent
1:!7 In This Video
1:25 Why Multi-Threaded Agents are Important
2:09 Two Verdent Version
2:45 Inside the Standalone Verdent App
3:15 Building a YouTube Script Writer in Verdent
6:40 7 Simultaneous Coding Tasks
7:32 App Build
8:07 Verdent in VSCode
10:50 Using Verdent
-------------------
Clips channel @jonlawedu
I do NOT sell any products of my own. Watch out for scams.
Queries: [email protected]
____________
Hey guys, so I was introduced recently to a very useful AI coding tool, and so I want
to share that tool with you in this video.
And my promise to you is that if you watch all the way through, you're going to get at
least one workflow or idea out of this that will materially make you more productive.
That in mind, the big problem with AI coding tools, we all know this, is that they're great
until you ask them to do something general or more than one thing at once.
And so the moment these very complex features or multifaceted prompts come into play, then
the builds just fall apart and you can't actually deploy, you just have to go back to square
one of troubleshooting, which is worse than the actual just build if you did it manually
most of the time.
Like, you'll fix a bug here, you'll start a feature there, you'll switch branches, you'll
ask it something else, and then the context is just gone, you get half-baked answers and
it just collapses in on itself.
And that's the idea behind this tool.
It's called Verdant, and it doesn't just respond to prompts, it runs parallel tasks while keeping
full context alive across all of those tasks.
Those are the big two selling points, and then the third for me is it plugs straight
into VS Code.
And so it's like if 10 agents were each working with one prompt on separate things, all with
complete context about the project, and plugged straight into your coding environment.
And so today I'm going to walk you through how exactly to use it, and some workflows
that you can actually implement after watching this video.
And to reiterate just why this is important, you know this if you code, or maybe you just
like vibe coding, there's always multiple things happening at once.
You're trying to work on this feature, you're fixing this bug, there's a whole laundry list
of things to do.
And yeah, you're going to focus on one thing when you're like just coding this feature,
for example, but maybe as part of doing that you have to add a secret, you have to refactor
something, you have to check some kind of logs, and when you ask AI to do that, now
it's trying to work on different things, and that's what kind of makes that oftentimes
that vibe coding collapse in on itself.
So that's why this kind of parallel multi-threaded approach to programming is important, because
it's what humans do, and now this tool is focused on helping AI do that as well.
And so Verdant comes in two versions, and number one is the standalone app, and this
is kind of the flagship experience that's designed for deep multi-tasking coding work.
And then you also have Verdant for VS Code, so that's the plan, code, and verify workflow
baked right into your standalone coding environment.
So they both use the underlying agent system.
If you already do everything in something like VS Code, then I'd say just tack it on
there.
If you're maybe newer to programming and you want more of a full-on vibe coding experience,
then the standalone app is definitely more comprehensive, in my opinion, for an AI-only
workflow.
That in mind, first let's jump into the standalone app.
So we'll go into my computer and I'll show you what that looks like.
Okay, so this right here is what the standalone Verdant app looks like.
Very intuitive overall.
Generally right here we can go ahead and switch the model.
You can see all of the leading, the cutting-edge models right in here.
We can see our projects, under projects, and then tasks when we're running an active project.
And let's start by asking Verdant to create a end-to-end YouTube script writing app, and
we'll see how it approaches these multiple different tasks.
Then I'm going under project, click open project.
And what we're going to start with is some kind of project that's already in our computer.
And so I'm going to go ahead and select this folder right here with some demo files.
And so the key difference we'll see here is that you're not limited to just one conversation
with the AI or task at a time.
We can open multiple conversations and pursue them simultaneously for this project.
So for example, first, what I'll say is create a user authentication system with login and
signup endpoints using express.js and JWT tokens.
Let's go ahead and start that task.
But now without waiting, so immediately while the agent is processing this task, I'm going
to go ahead and click new task.
We're going to reference the same project that we're already inside.
And then I'm going to ask it to design a responsive dashboard component in React with nav and
a staffs overview.
So we're going to start that.
Now we're opening a third task, we're going to ask it to write comprehensive API documentation.
Let's get that started.
And so we can run all of these in unison, right?
We have different cutting edge models all working together on these different requests
within our app.
And if we want to switch between them or see progress, you know, we just go right in here
and we can see exactly what is being done.
So we can see it went ahead and created a to do list and we can see all of those six
processes were completed there.
And it's working to make changes within the folder on my computer with the project.
So if we go scroll over into that folder in the computer, we can see a JSON file, some
client files and the API documentation.
And here's what that looks like.
This is extremely thorough.
And again, all of these are being pursued independently.
But what is very important here as well is that workspaces are isolated.
So you can experiment or try things within certain projects without affecting the code
from other ones.
And so what I can do is go to a new task now and I can enter this prompt to refactor the
off system to use OAuth 2 instead of JWT.
And I'm just saying, hey, this is experimental.
And so we're going to run this and you can see that as this is being done, this off system
and these files over here are not being affected.
So we can try something, see if it works, give it a feel if it does great.
If it doesn't, then no worries.
We can go back to our original setup over here.
And so it's the same thing with the dashboard.
We had a responsive dashboard that was created, but what I can ask Verdin to one of the agents
to do is implement that same component, but using Vue.js as opposed to React for a performance
comparison.
And what I can do by running this is I can create both dashboards simultaneously.
I can look at both working versions, I can test them and then choose my favorite and
remove the files for the other one.
And I can go ahead here, let's also just play around with this.
We'll go Opus 4.5 instead of Sonnet 4.5.
And you know, I think this kind of workflow makes a lot of sense when you see it here.
But when you think about what traditional vibe coding is like, you're having one of
these conversations at a time.
That's a pretty normal thing.
And again, going back to the core issue here, if you tried to ask Claude, just straight
in Claude to do all five of these things at the same time, the context window just wouldn't
be large enough.
It'd be trying to do everything and so it would fail at each individual thing.
So this system lets you break it up into all those requests, but we're still all working
cumulatively on that same aggregate project.
And to really illustrate that, let's fire off another five tasks right now to really
enhance this app demo.
This is going to fix a CORS error in the API, input validation on user-facing forms, unit
tests for middleware, and we're going very specific here.
This is optimizing database queries in the user lookup function, and now adding error
logging to API route.
So again, this is fully working.
We have seven different tasks, seven agents working simultaneously for us as we speak,
and we can switch between them at will.
Now, something I want to start to emphasize is the plan, code, and verify workflow that
these agents use.
And we're going to see that more so in the VS Code demo, but you can start to see that
here with the little to-do lists that are being followed.
And we get little notifications when tasks are being completed in the top right.
So now I'm going to go ahead and we can actually see what is built.
I'm going to open this index.html file in Chrome.
Okay, and here's what it looks like in that index.html file.
It's basically an AI tool platform for consumers.
So you can see this looks pretty good.
We can see all the different tools that are right there, pricing structure, some testimonials,
and we do have auth built out and a few other features.
But of course, a lot more work would have to be really put into this to make it actually
fully functional and ready to deploy.
But this is what we were able to do in five minutes of work.
So that in mind, this is the standalone app.
And now let's take this and let's shift over to VS Code and what it looks like in VS Code.
Now I've opened up a existing project here.
So we're in the standard interface for VS Code.
These are the files on the left.
And what you can see once you install the Verdant app is an interface that looks like
this.
And if you're not familiar with how to do that, you go to this extensions tab right
here, you just search Verdant, it will look like this, and then you just click install.
And so what this lets you do is same idea of using agents to run tasks simultaneously.
But while you maybe are a little more prioritized in that process, and AI is more of an assistant
here as opposed to really doing the work for you, if that's what you prefer, respect.
And I think that is what most very high level engineers for sure do with AI.
And we can see three different modes here, going back to plan, code, and verify.
The agent can actually go ahead and code, it can do things for you.
You can also just plan out certain features, what it would look like, or you can go ahead
and just chat and ask things, for example, about your code base.
Let's go ahead and click on plan, for example, and I'm going to say add rate limiting to
the off end points, and let's go ahead and see what that plan looks like.
So first, and I personally love this, this just little feature, especially it's going
to ask us a few questions about what we actually want.
Let's just fill this out.
We'll say node.js, we'll say in memory, login and register, and block IP, let's click submit.
And so it's previewed a plan, now let's ask it to generate the full plan.
So I'll hit full screen here, we can see what that plan looks like.
And this is thorough.
I mean, off the bat, you're going to notice the chart here, this is a full on image visualization
of what the process looks like to build out this feature.
And we can see the implementation tasks, task one, task two, task three, and it just keeps
on going for this whole report.
Now that's the plan part of it, if we want to shift into the agent actually building
this, we just click start building, and it's going to execute on that task in our files
for us.
Okay, so we can see the solution file rate limiter.js was written here, I went ahead
and clicked accept right there, and then it actually implemented it.
And that gets us to the start of the third component, which is verify.
So the Verdin agents work to recursively look at their work and make sure they didn't make
mistakes and review it for any kind of errors.
And that is that third component on the agentic coding workflow here.
So overall, that is how the standalone Verdin tap and VS code works.
And as far as the install goes, it's extremely easy, you have to Verdin.ai, if you can download
for VS code or download for Mac, that easy, you just Verdin.ai and go ahead and download
it with that limited time free trial.
Generally, this is for anyone who is doing coding programming that is complex enough
that you need to do multiple things at once, which is pretty much everyone.
And if you do just want to vibe code, then at least vibe code in a way that will get
things done.
And if you don't like vibe coding, then I think integrating an AI tool or assistant
into something like VS code or your current code editor is the best of both worlds.
It can help with debugging, it can help with small tasks.
And all in all, if you've made it this far in the video, I'm sure you can see the value
in this kind of tool for you.
So if you want to try this out, there's a link in the description and that does support
the channel.
And again, it's available as the Verdin standalone app or the Verdin for VS code like we looked
at.
And it is completely 100% free to download and experiment with Verdin.
You get a seven day trial and that includes frontier models, so top of the line AI model
requests.
So there is literally no risk for you to just give it a test spin and see how it fits into
your workflow.
So go hit that link in the description.
Again, no risk for you to try it and give it a go.
And that said, I wish you all a happy holiday season wherever you are in the world and I'll
see you in the next one.
Cheers.
