Everything you see in the first few seconds of this video was built by a single AI agent, one prompt each: a working web app, a full competitor analysis, and a pitch deck with charts. And that last one I kicked off just by talking to my phone. This isn't Claude Cowork, Cursor, or Replit. It's Trae Work, which just launched on mobile to complete a connected desktop, web, and mobile agent ecosystem. Most agent platforms that shipped over the past year were built for coders, so if you're a PM, a marketer, a founder, or a data analyst who doesn't write code, there hasn't been much for you. Trae Work is positioned squarely at that gap, with work mode handling documents, decks, research, and data analysis alongside a code mode that builds and deploys full stack apps. In this one I walk through the whole thing: real-time sync across all three surfaces, dispatching parallel tasks from your phone to multiple computers, native PowerPoint and Excel and PDF output, the skills marketplace, visual editing with comments on generated landing pages, scheduled automations, plugins, and auto model routing across frontier models. I run three live demos, including a voice input build where I describe an app idea out loud and the agent starts on it. Pricing, free trial options, and my honest read on who this is actually for are all in here too.
CHAPTERS 0:00 What One Agent Built With One Prompt Each 0:16 What is Trae Work? 0:36 The Voice-Enabled Desktop, Web, and Mobile Ecosystem 0:46 The Problem: Most AI Agents Are Built for Coders 1:05 Who This is For (PMs, Marketers, Founders, Analysts) 1:24 The Three Surfaces: Desktop, Web, and Mobile 1:33 Dispatching Parallel Tasks From Your Phone to Multiple Computers 1:45 "Think Anytime, Ship Everywhere" 1:55 Trae SOLO is Now Trae Work 2:02 Trae Work vs Claude Cowork 2:13 Native File Support (PowerPoint, Excel, PDF, Docs, HTML) 2:27 Universal File Management and the Workspace File Panel 2:43 Skills, Memories, Rules, Repo, and Wiki 2:54 How to Get Trae Work on Desktop, Web, and Mobile 3:14 Web App Walkthrough: Work Mode and Desktop Side by Side 3:40 Work Mode vs Code Mode Explained 4:04 Quick Look at Trae Design 4:14 Demo 1: AI Writing Tools Competitor Analysis Prompt 4:32 One Prompt, Three Deliverables (Deck, PDF, Landing Page) 4:52 Using Skills and the Trae Skills Marketplace 5:10 Watching the Agent Research and Build in Parallel 5:31 Reviewing the 10-Slide Generated Pitch Deck 5:48 Reviewing the Full Market Analysis Report 6:08 Reviewing the Generated Landing Page Concept 6:17 Visually Editing Elements Directly on the Page 6:31 Leaving Comments to Rewrite CTA Copy 6:45 Batching Multiple Comments Into One Prompt 7:06 My Tip: Make Edits in Bulk 7:25 Demo 2: Switching to Code Mode 7:35 Building a Subscription Tracker Web App 7:47 Deploying to GitHub vs Storing Locally 8:01 Working Out of a Local Folder 8:13 Setting Up Scheduled Automations and Agent Briefings 8:39 Plugins and Auto Model Routing Across Frontier Models 8:53 Trae Work Pricing, Free Credits, and the 7-Day Trial 9:15 The App is Built in Under 2 Minutes 9:38 Testing the Subscription Tracker Live 9:59 Requesting a Visual Fix in Code Mode 10:15 How Work Mode and Code Mode Complement Each Other 10:30 Demo 3: Voice Input on the Desktop App 10:53 Voice Input vs Voice Discussion 10:59 Talking an App Idea Into Existence 11:31 Trae Design and the Context Layer 11:51 Review: Real-Time Sync Across Every Device 12:08 Interrupted Tasks Continue on Your Other Devices 12:17 Review: The Layered Context System 12:32 Final Thoughts on the Trae Work Ecosystem 12:46 Who Should Actually Use This 13:01 Thanks for Watching
------------------- Clips channel @jonlawclips
I do NOT sell any products of my own. Watch out for scams.
Treywork is a cross-platform AI agent ecosystem designed for both technical and non-technical users. It features a synchronized experience across web, desktop, and mobile, offering specialized \"Work\" and \"Code\" modes to handle everything from business deliverables to full-stack app development, all enhanced by voice integration and a persistent context layer.
So everything that you are seeing right now was built by one AI agent and with one prompt each.
So a working web app, a competitor analysis, and a pitch deck with charts.
And the last one here I was able to kick off by just talking to my phone.
So it's not cloud co-work or cursor or replit.
It's a tool called Traywork and it recently launched on mobile as well to complete a
full desktop web and mobile interconnected agentic ecosystem.
So we're going to cover this ecosystem.
Take a look into it.
Now, a lot of what you saw isn't code.
That's going to be important to remember.
The other important thing is, again, they've now completed this voice-enabled ecosystem
across the web, the desktop, and the mobile touch point.
That's the video.
Let's go.
So the problem with a lot of AI agents that have dropped over the last year is they're
built for coders.
They ship software.
They ship websites.
They assume that you are fluent enough to be able to be at least comfortable in a dev
style workflow.
Now that's great.
In a lot of cases, we review a lot of those tools and the advancements and use cases on
this channel.
But if you're a PM, a marketer, a data analyst, you don't code there's room there.
And especially for someone who likes to get ideas or discuss things while they're talking.
So while they're just walking down the street, having coffee, right.
Or talking to a friend.
So Traywork kind of positions themselves as the solution to that.
So again, they have three surfaces, your desktop, your web, your mobile.
So you have your same agent, workspace, files, and contacts across those surfaces.
Now the product itself runs on the cloud, but you can also pair your phone with local
desktops.
And that lets one mobile device dispatch tasks in parallel to multiple computers.
And then also the cloud at the same time.
And their slogan is think anytime and ship everywhere.
I think that's the app descriptor where the industry is going and definitely something
that Tray does very well.
And a quick note as well.
This used to be called Tray Solo.
You may have heard of it in that name.
Now it is called formally Traywork.
Now if you think about this versus something like Cloud Cowork.
So Cowork kind of blends those into one, something like Tray more delineates them out.
Then second, you have broad native file type support on the Tray side.
So native docs files with tables of contents and citations, real PowerPoint, Excel, PDF,
HTML, that kind of full document stack.
And then third, there's universal file management on the Tray side.
So work as a file panel built into the workspace.
So session files are organized in one place.
And on desktop, it can natively, which is kind of the default across the whole ecosystem,
engage with files on your computer.
And then work has skills, memories, rules, a repo wiki, and external docs.
And then again, work is equivalent across all of the touchpoints, the web, desktop,
mobile.
With that in mind, let's get into the demos.
We'll jump into my computer.
If you want to follow along, you can do so at this link on the screen or in the description.
Desktop, web, and mobile are all on that site.
So if you want to run those cross platform workflows or experiments, you'll need to
install multiple of them.
And with that in mind, let's jump into my computer.
Okay, so here we are in the web version of Tray.
We can see we're in work mode.
I also have the desktop version pulled up right here.
It's essentially the same interface.
The mobile version looks the exact same as well.
And later, you can see exactly how you can interface and maintain a synchronous nature
across these touchpoints and platforms.
For now, let's just stay in the web version.
Now you can see Tray work right here.
That's basically everything except coding.
You want your PowerPoint, your data analysis, deep research, everything that you as a marketer,
a founder may need.
And then on the coding side, this is what we're all familiar with to build full stack
apps and deploy them.
We can connect to GitHub, etc.
Something new that we're not going to touch on in this video, but is pretty cool.
And you can see some of the demos here.
They look beautiful.
Is Tray design, but that won't be the focus of today's video.
We're going to stick mostly to Tray work.
And let's start by giving it this prompt.
So research the AI writing tools market and build a competitor analysis covering all these
various tools.
And then I'm just describing what that want.
And then I'm describing what I want just feature wise to understand about those different platforms.
And then you can see a number of different deliverables here we're getting from the
single prompt.
So a 10 slide pitch deck, a PDF, and then a landing page concept.
So we're getting a lot, a lot of work, right?
It's in the name out of one prompt.
And this is something we could ask on the mobile app or desktop or web.
Let's go ahead and click send.
And we can see it's using skills here.
That is something we already mentioned.
And you can check in.
And if we press on skills, we can see the whole marketplace of the different skills
here.
So that ranges from consulting, front end design, anything specific that you want your
agent to be really good at.
So we can see the steps here if we want to break it down.
So it's searching the web.
It's doing all of that research for us.
Now we can see in parallel, it's developing the report, the PDF and the landing page product
from the thesis that you can see it's developed here.
Okay.
And now it's actually building those files.
And here we go.
We can see all of the output.
Let's press on the deck.
And so here we can see our 10 slide fully generated deck.
And what's important here is the research layer.
And we can look here and we can actually see it's done its research, right?
The competitor map with what each of these does so we can understand it better.
Then let's go to the market analysis.
So here we can see more of the full stack report, the written report with the summary,
a comparison across all of the verticals we requested in the prompt patterns in the category.
And that continues here along with our listed sources.
And then finally, we have the landing page.
Let's extend that a little bit.
So here we can see the landing page concept.
Now what's really cool about this is this is a generated landing page.
But what we can do is click edit.
And then we can go ahead and select certain elements that we want to be changed.
And then we can either directly edit it like so.
So we can change the size, for example, and that's natively editing this.
Or we can go to the comment and we can say something like this.
Make this CTA copy more action oriented and urgent.
That is going to send this back to the chat, the chat portal here.
We can see the comment mark.
And so let's bundle a few of these.
Let's go here.
And then I will say reorder these bullets.
Now we have two comments.
And now I'll just say do this.
And it will execute both of those comments from our little edits on the page.
So really useful flow for updating visual content.
And that is what I recommend.
Do it in bulk, right?
Go through, make 10 changes at once.
And then in one prompt, you can get the whole project wrapped up.
Okay, here we can see those changes were made.
So we have our stronger headline here.
And then the order herein has been changed.
Now that overall is tray work mode.
Now what we can do is jump into code mode.
So we just go from work to code.
Now we're in this platform UI UX stack that is focused around code.
And so let's give this a test run.
Let's see what it can build.
So build a subscription tracker web app where you can add your recurring subscriptions.
Kind of a personal finance-esque tool like we talk about on this channel.
Now you can connect to GitHub to store and deploy that project.
In this case, we can just store it locally and test it out here.
Now what you can do, and this is important if we jump into the web version,
is work out of a folder.
So we can deploy locally.
We can store in the cloud.
Or we can go ahead, select a folder, and then store the code on our own computer.
Something else important to note before we trigger that prompt are automations.
These are extremely useful in line with the whole agentic trend.
Whereas agents do more work regularly for you.
That's things like a daily briefing, monitoring, anything.
It's just agents that can run on a schedule.
And you can create with these templates or in chat.
You can access them from your phone, web, or your desktop interfaces as well.
Final shout out here are the plugins.
That in mind, we're back here.
We're going to leave it under auto model.
You can see a lot of the frontier models that it can route through.
So it's going to optimize that cost and performance spectrum for you.
Let's let it run.
Now while this is working, here's what the pricing for the platform looks like.
If we go monthly, we can pay as low as $3 a month.
You can do that either without giving a credit card.
If you just want to sign up, give it a go with some test credits and a prompt.
Or you can try it free for seven days with a credit card with Pro.
And you can go up from there to pay.
But overall extremely affordable for what you're getting, especially compared to alternatives.
Here we can see it started developing the files.
So the CSS and the structure.
And there we go.
That took maybe two minutes, super fast.
It's done.
And if we click extend right here, we can see the subscription tracker.
We have the same editing functionality we did prior.
So very visual editing.
And then let's see if it works.
We can go, let's just go with the example.
Netflix for $19.99.
Billing dates and cancel by date.
Let's click save.
And that does successfully save right here.
And then we get our monthly spend projected annual and upcoming.
That's super cool.
And then again, if we want to make changes, make the font colors better.
It's hard to read.
And if we close this, we can go see it's working on that.
Now, overall, again, this is basically the code agentic capability that you're used to.
And then you get the addition of the work mode for everything else.
So they nicely complement each other.
And then there we go.
We can see that finished again, another super fast change.
So there we go.
Now, the last major point I want to touch on here is again, what I've been saying the whole time,
which is the different touch points that you have.
And then those accessory features that link them together.
So the fact that I can be on my phone in the mobile version of Trey,
and then I can just talk to it and then I can go build something throughout the rest of the stack.
And to give you a visual example of that, let's jump back into the desktop app.
And then right here, and you can see this everywhere on web and mobile as well.
I can click on voice input.
I can also click on voice discussion, which is an extended version of that.
But let's click on voice input.
Let's click allow.
Hey, Trey, I have an idea that I just talked about with a friend where we'll build a
mobile app and a web app that provides auto chapters for people's YouTube videos.
And for creators, can you go ahead and build this for me?
Let's click finish.
And we can see it put that in there.
And I can click send to get that kicked off.
So you can talk to the system again on any touch point, including mobile.
And there you go.
Overall, that is what I wanted to show you across this full stack of work mode and code
mode with Trey.
Now, final shout out again to design.
I'll leave that for you to explore on your own time.
Now, for that context layer, quick shout out to the plugins you can see here to connect
to different apps.
We also have the skills, the memories, the repo.
So in terms of a review and what I think matters most about this ecosystem, the first core
feature is the real time sync across all of those touch points.
So whatever you start on mobile shows up on the website instantly.
Or again, you can dispatch tasks from your phone to, for example, a bunch of agents on
your computer.
Same thing there with tasks that are interrupted.
If your phone dies or you delete the tab or something or close the app, whatever it is,
that task continues on those other devices.
And then number two, they have a very layered and built out context layer.
So the skills, the memories, the rules, the repo, et cetera.
The idea being that you kind of build work once and you don't have to re-prompt it.
It has that context and can improve itself over time.
So that is the full Trey work ecosystem, your desktop, your web, your mobile, and the one
agent that does everything across all of those screens.
Now, again, the link to try it out is here in the video or in the description.
If that's something you think would be useful for you.
And overall, I view it as most being useful for people who like to talk and have ideas
out there in the world, and then immediately get started on them across that whole suite
by talking to your phone, by just typing instructions on your laptop.
And overall, there you go.
That is Trey work.
Thank you all for watching and I'll see you in the next one.