Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 726: Perplexity Computer: What it is, How to use it and is it better than OpenClaw?

Episode Date: March 4, 2026

Spent hours testing AI models? ⏳Same. Openclaw setup still broken, heartbeat still dead? ☠️Yep, been there. Perplexity’s new agentic tool called Computer might’ve finally solved it. With no technical knowledge and a simple prompt, you can have agent swarms actually working for you and getting stuff done. Did we finally arrive in AI agent nirvana? Probably not. Because as impressive as it is, there’s one big caveat that will likely stop it dead in its tracks. Perplexity Computer: What it is, How to use it and is it better than OpenClaw? — An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:Perplexity Computer Hands-On Demo & ReviewPerplexity Computer vs. OpenClaw ComparisonPerplexity Computer's Multi-Model Agentic SwarmsAdvanced Workflow Automation With Perplexity ComputerAI Model Auto-Routing and Task ChainingData Integration: Gmail, Slack, Notion, Google DrivePerplexity Computer’s Price, Credits, and Value LimitationsBest Use Cases: Perplexity Computer vs. OpenClawCreating Animated Data Visualizations with PerplexityTimestamps:00:00 "Perplexity Computer Overview"03:08 AI Demos: Perplexity vs Big Four07:19 "Complex Queries with AI Tools"12:38 "Perplexity AI vs Open Claw"13:59 "AI Visual Analysis Overview"17:11 "AI Tools for Task Optimization"20:20 "Comparing Top AI Tools"24:51 "Perplexity vs. OpenClaw Comparison"27:18 Perplexity vs OpenClaw: Key Differences31:34 Daily AI Insights Workflow34:54 "Impressive AI Task Management"37:20 AI Tool Comparison and Limitations40:02 "Limited Value for High Price"Keywords: Perplexity Computer, Perplexity AI, Perplexity Computer features, Perplexity Computer vs OpenClaw, OpenClaw, AISend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the all-in-one creative AI studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Have you ever spent way too much time thinking about and testing what AI model is best for a certain job?

Or maybe you've already spent like six hours trying to get OpenClaught set up, but the telegram stream isn't telegramming and the heartbeat thing isn't heart beating? Yeah, me too. That's where I'm at right now. But I think Perplexity's new computer offering, may have kind of solved that. Here's the reality. AI agents and agentic models are finally getting their shine this year

and are already dominating the business landscape. But it's kind of overwhelming. I mean, there's so many agentic models. There's so many AI agents and different ways to connect them. The MCPs and the A2A's and the ACPs, it's AI agent soup and it still requires a ton of, of human oversight. That's why I've been pleasantly surprised so far with Perplexity's computer,

because it requires hardly no human duct tape and it has all of the best AI models and agentic swarms that just work and get work done for you. Yet, I still don't think I'd ultimately recommend Perplexity Computer right now for most of our listeners, and it's not for the reason that you're probably thinking. All right. Are you ready to put AI to work on Wednesdays? Let's get straight to it. So on today's show, here's what we're going to be going over. I'm going to show you what perplexity computer is and what it can unlock for even non-technical professionals.

We're going to go hands on. Yeah, live to showcase some of its advanced capabilities. We're going to compare it to one of its kind of closest competitors in OpenClaw. And I'm going to tell you if I think it's worth using or not. And yeah, spoiler alert, I think for most people, it's going to be a hard no, even though I am wildly impressed with its capabilities so far. All right. If you're new here, welcome. This is Everyday AI.

My name is Jordan Wilson. And while we do this every day, we do AI. It's a daily, unedited, unscripted, live stream podcast and free daily newsletter helping everyday business leaders like you and me learn and leverage AI to grow our companies and our careers. starts here with the unedited, unscripted podcast. But make sure to go to the next level. That's our website, your everyday AI.com. We're going to be recapping the highlights from today's show, as well as giving you all of the other AI news, you know, to stay ahead and be the smartest person in AI.

So let's talk about perplexity computer. All right. So on most Wednesdays, we do the putting AI at work on Wednesdays, where we go quite literally hands on. And then we do some demos and some, you know, really going behind the scenes on a new feature or a new AI model from one of, usually the big four. You know, usually we just focus on Microsoft co-pilot, chat, GPT, Google Gemini, and Anthropics, Claude. But I think this new offering from perplexity is good enough that we had to get our hand sturdy. And yes, for our newsletter audience that voted on our poll, everyone voted for Claude Co-Works, schedule.

tasks, which I was super stoked to do, spent hours, you know, trying to plan the show and get everything set up. But Anthropics been having some crazy outages right now. According to them, it's because they've had a surge of new users. So I was hoping to bring that show. Hopefully next week, I'll be able to use Anthropic Claude for more than 15 minutes without it breaking. But why not? Let's start live. What could go wrong? All right. So this is, perplexity's new computer. Unfortunately, right now, FYI, you do have to be a max plan subscriber. That is $200 a month, although perplexity did say that this is going to be coming out to pro subscribers, though on those on the $20 a month plan. All right, live stream audience, help me out.

Let me know if you can, you know, kind of see my screen here. Maybe zoom in a little bit. All right, podcast audience. If you ever want the video version of this, it's always available on our website at your everyday AI.com, but I'm going to do my best to describe this to you. So I've only been playing around with computer for a couple hours here, you know, just really trying to get my, my hand sturdy, you know, trying to really push it to its limits, just in a, well, very limited capacity, though, and I'll tell you why here in a minute.

So here's how it's going to work. You're going to go to your normal perplexity plan and you're going to see a new option, right, where you would normally just type in a prompt and, you know, choose your model or your mode inside of perplexity. Now there's a new option that says computer. All right. So I have something I wanted to try. All right. So I'm going to click computer.

And this part is a little different. And at first, I wasn't a big fan of this interface. But after a while, well, I kind of like it. So I just said, create a bar race chart. over the last five years showing monthly prices for NVIDIA, Microsoft, Tesla, Google, Apple, Amazon, and meta. So the MAG 7AI stocks. So if you've ever seen those kind of race, I think they're called bar race charts, right? But it's essentially where it plots different stocks or, well, anything that can be plotted on a spreadsheet on a on a graph, on a bar chart,

right? And it shows you kind of something over time. So what, I don't know if this is going to work. If it does work, we'll, you know, maybe post it on Twitter and post in the newsletter. I don't know if this will actually work because it's a kind of complex task, right? So think of, you know, as we go into a little bit here on what a perplexity computer is and some different ways to use it, I want you to think of some of those very difficult knowledge work tasks like you do, right, or that you do on an ongoing basis. And that's one of the reasons why I thought of something like this. I've always wanted to create these and have them as supplemental information or

complimentary information that goes along in our newsletter. But things like this are hard, right? I used to edit videos 10 years ago, but I've spent thousands of hours doing it. There's some kind of AI tools that can help do these things. But as far as I know right now, there's no easy, simple way to make something like this. All right. So I'm going to show you right now, kind of what's going on and walk you through it. So the way that I would normally do this right now, I don't know, right? I would probably do a a GBT 52 Pro query to figure out the best way. I would still try doing something like this

in Opus 4-6 GBT 52 Pro and, you know, Gemini 3.3. I don't think any of them could necessarily do this because it's very complex. I don't even know if this is going to work and if it does work what the end goal is going to be. Is it going to be a GIF and animation? Is it going to be a video? I don't even know to tell you the truth. If perplexity computer can generate video or not.

So we're going to find out as we go along. But you'll see right here on the right hand side, there is this chain of thought. So kind of the new layout that it took me a little while to get used to is you now have a three pain layout when you're using computer. All right. So if you use notebook L.LM, you know, maybe this is a little easier on the eyes. And you can toggle the right hand column and the middle column to get a different size. So essentially what you have here is you have your, you can always go back to your normal perplexity view. So if you are a perplexity user, that didn't change. But then in the computer view after you've entered into a certain task is what it's called, you're going to see

all of your other tasks there at the same time. So as I was testing this, now I have four on the screen. I was running three concurrently. I was seeing, okay, is it going to break? Are we going to go off the rails if we're running too many tasks at once? And it actually did a pretty good job. And it built these in a pretty quick amount of time. So let me just walk you through kind of what's going on here. So right now it says, I'll load the relevant skills for this task and gather the stock price data. So now it's running this task in parallel. It's loading different skills, right? So it has different skills that are already built in. I didn't go out and build these skills, right? So obviously skills popularized and

brought to market by Anthropic, but they've been adopted by, you know, everyone from Microsoft and Google to Open AI, right? Everyone's using skills. So it's using these skills to do some research. It looks like it said, let me get the historical price data. Okay. So as I go through here, you know, I can click around and see what it's doing. It looks like it's running a little bit of code, running some JSON, pulling some API responses. And you'll see right now, wait, is it done? It might be done.

Let's see. It said the bar. Yeah, it just finished. It says the bar race chart is live and looking clean. It shows all seven tech stocks with monthly closing prices from March. March 2021 to March 26 using data from Perplexity Finance. Okay, that's one of the reasons why it was able to go so fast. I forgot perplexity has a really good and really popular finance tool.

So it says hit play to watch the race unfold. So let's go ahead and see this. I don't think this is going to generate exactly what I wanted. But not bad. Okay. So what we have here is we have actual bar chart. So I should have done a little bit better job in the prompt of describing this that I actually wanted to see kind of more on the candlestick line, right, then go up and down. But instead, I mean, we'll see.

So it has kind of the standard bar chart. So right now, Nvidia's at the bottom, right? Because this is five years ago. And this is, I'm trying to see what our access is here. I'm guessing it's billions. Yeah, it looks like it's billions in market cap. I'm guessing is what it is because it has met. at the top at the time, Microsoft, this might be revenue.

I'm not sure. So, oh, no, it says monthly closing prices. So that's not great either. I'm probably going to actually play this for everyone. So at least our live stream audience can see. I'm going to actually go in and edit this prompt a little bit. And then we're going to look at some of the other results. So let's go ahead and play.

OK. So to their credit, it is animating and it's doing fairly well. So when a new company kind of goes on top, it's switching their position. So as an example, NVIDIA, okay, I see what's kind of happening here. NVIDIA is not going to climb all the way to the top because they did have their stock split. Right. So I think it went from like, you know, $1,300 down to 130. And it looks like it's retroactively adjusting it.

So this is just doing it by stock price, which isn't necessarily a good barometer to see which company is growing the most. But like I said, to their credit, I didn't really denote that in my prompt. So I'm going to go in here and, well, we'll just go ahead and... Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the All In One Creative AI Studio. Powered by Adobe's Creative Agent, Firefly AI Assistant lets you start with your vision, just describe

what you want, and shape the outcome as it takes form with the Assistant. The assistant orchestrates multi-step workflows drawing on 60-plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect or take over at any time. You stay in the driver's seat as the creative director.

Adobe Firefly AI assistant now in public beta. See it today at firefly.adobie.com. See how to do it live. So I'm going to say let's update this to market cap so we can see a truer version of company growth over the years. All right. We're going to do nothing else. we're going to come back and check on that.

So let's go ahead and, well, this is going to be maybe a fun one. Because I asked perplexity to do my work for me. All right. So in this prompt example here, I simple. I said, hey, I'm doing this live stream. Go do my presentation. So I said, create a balanced, insightful and factual presentation that I can show for my podcast every day AI.

carefully find and emulate the exact style of my live stream presentations and create a presentation for this podcast, this very one I'm doing right now, called Perplexity Computer, what it is, how to use it, and is it better than OpenClaw? I said, please do all the research, organization, and design. So you'll see here again, it chooses which model to use, which is great. So it's loading a different skill. So it loaded an office PowerPoint skill. And then it said it's running these different tasks in parallel. So it's kind of putting out different subagents and it's deciding what these subagents are going to do. One thing I wish that you could dive into a little bit more depth on how some of these skills are used.

But it did. I can click and it's showing me, you know, what it's searching on the web. You know, what is Everyday AI podcast presentation style? It says, I don't have any previous records of your presentation style. Let me find examples of your Everyday AI. live stream. So it went through. It said it created task list here. So pretty cool. It's running these tasks in parallel. And then it said, okay, I think I have a solid understanding of this topic. Go through here. Let me just keep browsing a little bit. And then you'll see here we're starting to use

different models here. Right. So now it's using Gemini 3 Flash to analyze visuals, which is pretty cool because it used a, I think a Python tool to take a screenshot. It found a YouTube presentation, took a screenshot, and then it said, okay, I need to use an AI model to understand this visual, right? Again, think of all the human work that's something like this would normally take, right? So it went and it used Gemini 3 Flash to kind of understand and ingest a screenshot of our visual. So then I went from there and it took more screenshots. of the live stream presentation. All right.

And then, yeah, so it literally says right here. I can see what video it used. It looks like it was a video from last week. It said it's scrubbed to around three to five minutes where the video presentation slides are visible on the screen. It's taking these different screenshots, throwing them back into Gemini 3 Flash, right? So pretty cool just to look under the hood and see how perplexity computer works. Right. And then it kind of started to pick up on some of the branding.

it's, you know, dark navy, background, a turquoise accent color. I actually got the color code correct, white text for primary content, bold, San Serra for headlines, right? So it did a really good job. I didn't have to tell it. Here's my brand guidelines, you know, et cetera. And then it obviously went out. It did the research as well. And then it created a presentation. So I did already look at the first version. And the first version wasn't the best. Okay. It was okay. There's nothing wrong with it per se. It just didn't really hit the exact branding that I would want, although it was fairly close. It looks like something that maybe I could, don't even wrong, it looks better than my normal live stream slides, right?

I say they're the ugliest things ever just because I put them together myself. So these actually look a little bit better, but they don't really look fully on brand. All right. So I actually go and repripping. So at any point, you don't have to wait for this to finish. At any point, you can actually send a prompt to computer to have it update. So essentially I said, this isn't really anything like my live stream slides. Please update the visuals and try to create a one-to-one replica of my standard slide setup for live streams.

And then I also kind of wanted to adjust a little bit on the content. So I haven't seen the results yet. So I'm kind of scrolling through here on the chain of thought. Again, we have the three panes. The left pane is your normal. normal navigation where you can go to the normal perplexity search, computer, your tasks, your files, your connectors, etc. In the middle, that's your task list where you can see the different tasks if you have multiple going on. And then on the right hand side, which is kind

of what I'm reading through and what I'm scrolling through now here on the live stream version, this is kind of your chain of thought. This is where you can see what these different AI models are doing. So let's just go ahead and go to the bottom here. Let's see if it did any better, you know, for our normal live stream audience, you'll have to let me know, you know, how close this is. So a ton, a ton of information. I'm seeing all the different tools that use, you know, it switched over here to Claude Sonnet 46 for some visual QA, right? So it used some models to build it, different models to research, different models to do QA after. Very cool. What this also shows to me is just how steerable this can be. If you do,

want to steer it, right? The great thing so far that I like about perplexity is it just uses the right AI model for the right job, even if you don't even know what the right model is or what the right job is, right? This is a lot of tasks. This is my example is extremely easy, uh, to follow, right? But that's why hopefully it'll encourage you or maybe inspire you to say, okay, maybe I could use a perplexity computer for workflow a. All right. So let's go ahead see. And then we're going to actually flip through these slides, right? Okay, here we go. All right.

So actually, much better. All right. So it actually did a much better job on the second iteration of sticking to our brand, right? It has the normal kind of turquoise heading that we use, the white text, the bottom bar with our website,

the top bar with the episode name, So it did pretty good. Obviously, it didn't pull the actual logo. That would be one of the main things missing. Otherwise, if I'm looking at this quickly, it looks pretty much like exactly what we put out there. Aside from a title slide and putting our logo on there, it actually did pretty, pretty impressive job. All right. But I'm going to, I'm going to actually go up and use this because I did look at the first one and I'm like, this is great. And for our podcast audience, right? I think some of this information in there. is going to be really important as we learn a little bit more about perplexity computer. So one of the most important things to call out, I think here is 19 models. Okay. That's how many models that perplexity computer right now can use in parallel. All right. And like I said, all you need to do is just describe the project.

You don't have to be super technical. You don't have to, you know, be super advanced in prompt engineering. And then computer breaks it down. into smaller tax and then it routes to the best AI model. So tons of different models that it can use. Okay, so it does have video generation. I actually knew that. I just had temporarily forgotten it, right?

I need longer context window. I need that three million token context window in my own brain. So, you know, for reasoning as an example, it can use Claude Obis 6. For research, it can use Gemini 31 Pro. For image generation, it can use nanobanana. For video generation, it can use V031 for long context. I can use GBT 52. For fast tasks, it can use GROC, right?

So for like this to me, this is so much of what I do on the day-to-day basis. Even though I do think ChadGBT is probably my most used AI system. And then, you know, Claude and Gemini are kind of tied there for second most use. I think if I did more creative things, I would probably use Google Gemini way more just because Nanobanana Pro and B031. But this is a lot of what I would be doing on a day-to-day basis, right? Going between these different AI tools, these different models, I have subscriptions to all of them and I use all of them.

So this right here, this kind of engine that computer runs off of is extremely impressive, especially if you are, well, number one, I think what this might appeal to is people that know they want more out of AI models, but maybe they don't want to subscribe to all of them, this does kind of give you the best of all worlds without you having to even know, which I think is actually really nice to have. Because unless you're a dork like me and can keep up with it, you probably don't know which AI model right now ranks best for grounded search. Because if you read the newsletter last week, you would know it was Gemini 3.1.

But no longer. Now it's opus 4.6 on web grounding. Right. So it's it's really difficult to know. which AI model or which mode is the best for the right task. So that's why using something like computer is extremely helpful. So some of the key capabilities, well, you can connect your data, your integrations, your Gmail, your Slack, your Notion, Google Drive, and more.

And it does have persistent memory. So it can remember past work, preferences, and context across sessions. Like I already talked about, you can spawn specialized agents without even knowing or without having to do anything. Right. And then also it runs in an isolated environment. That is one of the big differentiators between something like perplexity computer and open claw. So what the heck can you do with perplexity computer? Well, a lot. It's almost like, well, what can't you do? Right. So deep research with source synthesis, code projects from spec to deployment, content pipelines, competitive analysis. You know, really any type of knowledge work. Because like I said, video,

photo, infographics, research, it's all in there. So one narrative that I saw when perplexity computer was announced

and that I kind of agreed with because I'm like, okay, what is the closest competitor here to computer? At first, I'm like, okay,

maybe it's Manus, maybe it's GenSpark, it's more of these, you know, it's a super agent. That's what it is. Right. But the more I thought

about it and looked at it, I'm like, this seems, seems just maybe more like a more enterprise ready and secure version of open claw. Granted, probably more expensive, right? And I say probably because OpenClaw, if you don't know what it is, well, it's a free open source project. It's an autonomous agent that can run on your local machine.

If you give it access to it, you can. You can run it in a cloud environment as well. But what a lot of people are doing is they're giving OpenClaught its own. you know, Mac Mini, Mac Studio, whatever, sometimes downloading local models to save on costs, but a lot of people are still connecting it via the API to, you know, different GPT models, you know, different Claude models, you know, mini Macs, etc. So even though OpenClaught is free and open source, if you really want Frontier performance, unless you have a $30,000, you know, Mac Studio.

you know, you're still going to have to pay for it on the API side. So that's when I thought about that. You know, maybe in five years when you have frontier level models that can run locally, but at least right now, if I had to choose, right, if I could have technically unlimited, either perplexity computer or unlimited open claw, I would probably have unlimited perplexity computer because it is more secure. It's done for you. I've already spent hours.

hours, you know, trying to set up my open claw. That's why I haven't done a show on it yet, because it's like you need so much time to get it set up, right? There's updates every day. I mean, the team is amazing there and how quickly they ship things. But, you know, to go from zero to one, it actually takes a lot of time. Whereas with perplexity computer, it takes no time. You can put in the simplest prompt and it gets going. So, you know, the big difference with, you know, Well, there's a lot of differences between perplexity computer and OpenClaw. But even though at first the comparison, I was like, I don't know if that's comparing apples and oranges or if it's comparing apples and banana chips. But I think it is pretty similar.

But OpenClaw is more of a self-hosted AI agent that can run on your machine. And you can talk to it via WhatsApp, telegram, Discord, Slack, I message, etc. Right. The only difference, I think ultimately, because with all of those connectors, you can put just about any of that information into perplexity. complexity computer and its ability to have memory, a consistent memory across sessions. So yeah, I can't, you know, necessarily text your, your, your, your wife, or it can't, you know, necessarily, you know, drag and drop files onto your computer. But it can do just about anything if you are using all of those connectors.

So here's a little bit of head to head. So the architecture, perplexity computer, cloud-based sandbox, open claw, self-hosted. Set up, like I said, perplexity computer, nothing. OpenClaw, decent amount. Although you can, you know, get OpenClaw up and running in, you know, 10, 15 minutes. But if you want to do it the right way, it can take many hours. The cost, that's where it gets, you know, some differences.

So OpenClaw technically is free. If you do have a more powerful, you know, Mac Studio or, you know, Nvidia, DGX Spark, something like that, you can run. some of these, you know, like a Quen 3-5, you know, their new small models, right? But you still can't run with OpenClaw. You can't run anything at the level of an Opus 4-6, a GPD-5-2, right? You can't. You know, and if you chain five of them together, you know, you can do it very slowly,

but at that point, you're spending $25,000. You know, but really where I think the difference comes is just the, ability to set up, right? And even if you look at AI models, the fact that 19 models inside a perplexity computer are just auto-routed and you don't have to do anything, that's a huge advantage. All right. So at least if you're deciding between perplexity and computer and open claw, who is it best for?

I think right now perplexity computer probably best for business professionals and executives, non-technical people, people who need a little bit more security and people who just want to see and experience or actually use. agentics forms without having to spend a lot of time learning, tweaking, improving. OpenClaw, probably a little bit more for developers and technical power users. Or if you are ultimately more budget conscious. Because like I said, right now, perplexity computer, you do have to be on that $200 a month. But if I'm being honest, if you really want to get the most out of OpenClaw, you're probably

going to be spending at least a couple hundred dollars a month anyway on a API cost. So, you know, tomato, tomato. All right. So that's enough on the comparison. I do want to show you a couple other results of what we did inside of perplexity computer. I did actually see while we were doing that. It did say that the updated version was done of the bar chart. So let's see. Okay, this makes a lot more sense now because now to start with Apple is at the top and we do have market cap there. So, you know, here we are starting in 20, 21. And, you know, it's nice and it's animated there in the upper right hand corner says 2021. We have Apple with the top market cap at $2 trillion. And then Nvidia at the very bottom with a market cap of $332 billion.

So I'll play this. This is fun to watch. And it's kind of interesting for me, right, as Google, Microsoft and Apple in 2022 are kind of tossling for the the top spot. Invidia still kind of near the bottom passing meta. You know, Amazon's just kind of been hanging out

in the middle there through 2023, right? Obviously things with you know, COVID coming out of the severe COVID period. And now you just have Nvidia by 2025 all the way at the top, right? With a, you know, four, five trillion dollar market cap and huge, right? So that was kind of fun to watch. You know,

live stream audience. Let me know. Should I post this in the in the newsletter? I don't know if anyone cares or not, but at least for me, I always wanted to see, you know, this exact thing. I'm like, this is fascinating. And I've never been able to see it. And perplexity computer just did it for me with two prompts. And probably could have been one. I didn't do a good enough job describing it. So I do want to show you one or two other things because yeah, what you can build in here, literally anything. So in simple prompts too. So in this example, I just said, create a gamified task tracker for Mac. All right. So it went through. I can quickly kind of show it, use different skills. It said, okay, for this, Claude Opus 4-6 is going to be

the best. And it was done. And it told me that I can download this and install it. But it also said, Here's a web version as well. So I'm opening it here on my screen and it's a fully working gamified testing. Right. So I can say, you know, publish the perplexity computer podcast. All right. And then there's different toggles here for a one, a two or three. I'm guessing that's, you know, how much time.

Yeah, how much time you need. And then a personal or work toggle. I click enter. So it's a working task list, right? I can go check it off. There you go. And it looks like there's some badges. You know, it's a little gamified thing called Quest vlog.

So again, working. Very nice, very slick. I can go, you know, update it. I can create a database. I could tell it, go create a database, you know, sign me up for whatever I need. Or, you know, I'll take over the computer and sign up for whatever I need. All right. And then one last one.

And this is one. I always do a version of this. when I'm testing out agentic models. Because the agentic models themselves actually do a pretty good job of this. So I essentially have a saved prompt that I use and I have a in chat GPT agent mode. I have to do this every single day

and I usually start my day by reading something like this so I can keep up on the latest in a way that helps me. So I essentially say, here's a bunch of information about everyday AI. Here's the things that we care about. Here's the things that our audience cares about. I give it some URLs, some things to search, some reasoning, right? Here's reasons why certain stories matter to our audience more than ever, right?

And then I have it build an interactive HTML website, you know. So nothing crazy here, right? So simple, well, not as simple. It's a long prompt, but a simple prompt. And at the end, this is pretty impressive, right? It put together a sortable daily briefing. So this is the March 3rd, top 20 AI stories you need to know. All right.

I can sort by all. It put them in order. What's crazy is I'm saying like, yeah, this is going to be the order of our newsletter. I can sort by Google, by Anthropic, by invidia, by Amazon, right? So pretty cool. There's a search bar. Let's see if it works. I'm going to type in 5.3.

There you go. because it's opening I released GBT53 instant so it all works right I can click on it it gave me bullet points and I'm reading these these are factual so I'm not seeing any

hallucinations there's an update to the story that's pretty cool so again extremely extremely impressive with this one shot I mean look at this there's even a light mode and a dark mode I didn't even ask for

that. This is really good. When you talk about the future of personalized knowledge, consumption, if you're not doing something like this almost every day, you are getting behind. I'm tired of saying that, but I always show something like this and people are like, oh my gosh, I'm losing my mind. It's like, you need to be doing this, right? Everyone should be doing something like this, just a way to personalize and consume information in a way that is going to resonate more with you, but also think about how you can use it for your company, your department, your customers, et cetera, right? So that's it.

That's a wrap. Let me give you some of my final thoughts here. All right. So my big takeaways, the capabilities themselves extremely impressive. Technically, it is, when I was just watching at work and looking at some of the first outputs without really learning it, right? I was very impressed, right? I will say probably, and this might not seem like a lot,

but I've used, you know, I've been doing this everyday AI thing for three years. I've used probably at least, I don't know, 500 plus AI tools. This was a top 10, like, wow moment for me, right? Which is hard, right? When you think about things like, like VO3 and, You know, SORA and, you know, notebook Alam and nanobanana, right? To say like this is a top 10 thing, that's pretty big, right?

Seeing how seamlessly it worked and how you didn't even need to have those technical skills. And I intentionally gave it very vague prompts in two of the four cases just to see how good it could do. It is extremely impressive. Its ability to spin up subagents, assign the correct model to the right task, right? Even in there, I'm pretty picky about like using the right model. And it did really, really good on using the right model for the right task. Right. Even the example, right, it used Opus 46 for the coding task, right?

It could have in theory used Opus, you know, four or five. Could you use Saunit 46, but it said, okay, no for this. Opus 46 is the right tool for the job, right? So that's one thing. Very little human skill needed. The thing I talk about human duct tape and why the future of multi-agent swarms is so important, well, you just kind of saw or heard examples of why that is, right? Especially when those swarms can pull in models from different models and modes from different frontier AI labs.

Connections, you've got to talk about connections. So it just has, it integrates with platforms like Gmail, Slack, Notion, GitHub, Salesforce notion, right? So it can monitor and schedule as well, right? I should have mentioned that earlier, because you can run these things every day, every hour, every week, every month, right? So if you need the inbox calendar triage, right,

you can run that each and every morning, you know, grabbing information from your Google Drive and pairing it up with something from Slack or Notion, it can do it all, right? It can build you literally. I've seen examples of this. It can build you an update, keep an updated version of a personal CRM. Right.

It can use the browser. So so many the capabilities just because of the connections are huge. And I think that's where like a lot of the narrative early on was like, oh, no, well, I need open claw, right? Because I need to be able to give it all of this important data. Well, when an agent can, number one, browse the web willingly, but also, if you can connect your Gmail Slack notion, all those things where all of your data lives, aside from it being able to use iMessage on its own or, you know, things like that, I at least think that perplexity computer can do two thirds of the things that OpenClaught can do

with zero setup, right? Or one-click setup, bringing in your connectors. So what's this best for now? I think right now it's best for tasks that would normally require multiple models, multiple modes, or separate AI tools that would require that human duct tape between systems. So as an example, right, if I needed to do some personalized research and create a spreadsheet and a PowerPoint, I wouldn't use perplexity computer because I know Claude and chat GPT can do that in one shot and it already has my memory. And then this takes me to, well, the big thing, though, with perplexity computers just works. It's very impressive. But here's why I ultimately probably won't personally use this a ton and why I don't think at least now most of you should either. All right.

I'm going to hit refresh on my page here. Okay. So I just did those four little tests, right? Nothing crazy. with a $200 a month plan, you get 10,000 credits. Okay, so this is a credit-based system. In my little test there, I went through about 4,000 credits. So I went through 40% in an hour or so of, and I think a total, I hit the interkey,

I think a total of seven times. So it wasn't like I was iterating a ton. All right? That's probably not good. It's probably not good, right? Because the reality, you could say, okay, well, yeah, Jordan, it's worth it. It uses all these models, right? I do think especially between Open AI, Anthropic, and Google,

I think that they're going to have something comparable to this fairly soon. And from a value perspective, I'm just not seeing it because the credits burn so quickly. And to pay $200 a month, even that example, right? At least the nice thing is, is I can go in and see, you know, how much each of these things use. So that little newsletter example that I showed you, right, went through, did some research, put together a little HTML page, right? That took 800 credits. So if I'm doing the math, I can do that, what, 12, 12 days a month.

I can't even do that every day. So I can't even use that as my daily digest planner. Right. So I think the capabilities are extremely impressive. the value, I think for the most part, just isn't there. Granted, you know, perplexity does have a little promotion going on where you get some one-time bonus credits. But outside of that, I don't see this at least right now being worth that $200 a month, right?

Because at least for me still, I'd rather spend that $200 with Chad GPT. I'd rather spend that $200, you know, with Claude. I'd rather spend the $200 on, you know, Gemini Ultra. because at least with Google Gemini and Open AIs, the limits are almost limitless, right? So I almost felt very limited after I ran my first prompt and I'm like, oh my gosh, this ate through a thousand credits, right? So I'm like, oh, I can only use this 40 times a month.

I'm using AI all day. So at least for me, and I think for a lot of our audience, although I really want to love perplexity computer. And I do think it is better than OpenClaw in. so many ways. The value, just not there for me right now. All right. I hope this was helpful putting AI to work on Wednesday.

So now you know perplexity computer, what it is, how to use it. And if it's better than OpenClaw, at least for your use case. So if this was helpful, tell someone about it. But if this was helpful, you're really going to find our 2026 AI prediction and roadmap series very helpful. So make sure if you haven't already listened to that, episode 7, 12, and 713. Thank you for tuning in. Make sure to go to your everyday AI.com. Sign up for the free daily newsletter. See you back tomorrow and every day for more everyday AI. Thanks y'all. Meet

Firefly AI assistant. Now live in Adobe Firefly, the Allman One creative AI studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't

get left behind. Go break some barriers and we'll see you next time.