Anthropic Made a Safer OpenClaw & More AI News You Can Use

Alright, welcome to another week in generative AI. And yep, I'm fully recovered from my surgery. Thanks to AI Igor Pogany for hopping in and recording last week's episode. We'll have a little segment on behind the scenes on how we did it. But this week, there's fewer updates than usual, but there is a clear theme. Everybody's trying to build a consumer-friendly open clause. So we'll talk about that. Plus Google Maps is getting AI integrations. And there's an unbelievably interesting and cool feature inside of Photoshop I want to show you. All of that and more in this week's episode of AI News You Can Use, the show that pulls together all the generative AI releases and we filter by the things that actually matter to you as a non-technical consumer. Let's start with the big one because all stories recently out of the big companies and also the medium-sized startups, everybody is trying to build their version of OpenClaw. Now before I get into showing you what updates came out and what this actually looks like in practice, quick recap on what OpenClaw is. So If you're familiar with ChatGPT, basically a chatbot that acts as a work assistant, OpenClaw is a personal agent that doesn't just assist you with work, but it also gets things done, or it at least tries to get things done. People are experimenting with this stuff. It's the most overhyped thing on the internet. But as we've talked about on the channel before, there is some substance there. Now, OpenClaw requires its own machine. That's why a lot of people buy Mac Minis so it can run 24/7. That has access to the whole operating system. But some of these other apps that we'll be looking at now, they're trying to get around that. They're trying to make it more consumer-friendly where it either runs just through the browser, you just go to a site and use it there, you don't need an own computer. And there's versions of it that are actually closer to OpenClaw, like Claude Cowork. Now, this is where I want to start because I've been sharing some updates in Claude Cowork over the past weeks, just kind of on the side and saying, hey, they shipped this feature that brings it closer to OpenClaw and this one and this one. But really, when I go from using and experimenting OpenClaw to working with COWORK, there were two big things on my laundry list that I was missing. First of all, it was the ability to schedule certain tasks. So if you get something done once, the whole idea is not having to do it again. So scheduling it out, they added this a few weeks ago. You can see the schedule tab in here. You can do that. And the second big feature is it being a worker of its own that I can tell what to do at any point in time. And that's a few things. One of them is the ability to remote control it. And the second one is the base setup of an actual computer being there and available 24/7. The big missing link there was the remote control. You didn't have an ability to kind of just dial in and give it tasks from your phone where then all of a sudden it would move around files on your computer. Well, they just added that. And Froppy added this feature. They call it Dispatch. And look, it's available right here. And I'm going to give it a first shot live here on the show. You can see this is how they describe it. They say you can use your phone like a walkie-talkie that can communicate with Claude on your computer. Just remember to keep the computer on and running so there's something to remote control. I'm just going to scan this QR code to pair up my mobile app that I also need to update. So yeah, I needed to update this desktop app to have this tab, and I also updated the mobile app. And if this is going to be the remote control, the second thing is exactly here. This is what was missing: keeping the computer awake. I can just turn it on here and say finish setup. I have to give it access to my desktop folder, my documents folder, and there it is. Now I see the same screen on my computer and on my mobile phone. So let's do the classic use case that everybody champions as, oh, this is the big thing that all of these AI agents can do for you. Organize my desktop. And yeah, I sent it here and you can see it appearing on the screen. By the way, we're doing a fun little experiment. As I put up that TV behind me, we're mirroring the computer screen onto there. Not sure if that helps or makes the scene better. If you have any thoughts on this, you can leave them below. Anyway, I'll do all of this from my phone. It's asking me for access to my desktop. So I'll say allow once and I'm literally giving it access to my local MacBook that is you know, plugged into power here. That's recording this video. But also Cowork is, you know, getting some work done on this computer. Honestly, this is the last big missing piece. Sure, there's a few smaller things that OpenClaw still has. But this was the big one, like the ability to remote control it. And once you get something done, scheduling it out, it has both of those now. It made a plan. I'll just say yes, do it. By the way, beautiful interface. Also the loading bar background, everything just Looks neat. And there it is. It organized it. Let's have a look. Yep, sure enough, when I look at my desktop, it sorted stuff. And that's the gist of it. So I think this is really interesting for many reasons, but mainly because this is a way more secure version of OpenClaw that doesn't require a separate machine. And it's actually sort of polished. This is an Anthropic Claude product, not something that you have to install through the terminal. And it doesn't even need access to anything. It just wanted file access to my desktop and downloads folder here. You could even constrain it to its own folder so that it doesn't affect the rest of your system. We're word of warning, if you give it access to files on your desktop, I would always make a backup. Worst come to worst, give it access to delete those files. And there's a non-zero percent chance that that will happen somehow if you're careless with it. Which the more you use this, the more careless you get, because it kind of just gets things done. And you're also in this mindset of like, hey, this is supposed to do things for me. So just the very realistic tendency is not to double-check every single thing it does for you. You kind of just say, yeah, just go, just work. And mistakes happen. So there you go. And Propic is doing it. They're building the first version of this that is actually consumer-friendly. And I think I was a bit hesitant on covering OpenClaw on this channel because I think for most consumers, this was not the big unlock that a lot of people were making it up to be. There's value there, there's substance there. I think I'll create more content soon outlining how I'm using it and doing some tutorials on this type of stuff too. But I think I'll be focusing on this because this is accessible to most people. And I love that. Now, as I mentioned, there's a few more releases relating to this this week, which I want to touch on here. Namely, there's free. GenSpark is making their own version of Genspark Claw. Manus is introducing My Computer, which is essentially a competing product to Genspark Claw, which is Perplexity version of OpenClaw. The Perplexity one being by far the most polished version here when it comes to the interface. It looks very similar to Claude Cowork. Also, you need a desktop app, whereas Genspark and Manus run in the web. Now here's the thing though, I think even the makers of this are sort of confused by what the actual use cases of this are for most people. If you have a few minutes to spare, actually watch this video that Genspark put out titled "We Let Genspark Claw Run Our Company for a Day," basically being like a use case video from them internally. Guys, my conclusion from this is they're equally as confused as every consumer on what this should actually be used for. Like seriously, I'm not gonna go into more detail here, but it's like scheduling a family call for this lady and ordering a snack while it's setting off a confetti cannon for this guy while reading his email. And then it prints a ticket to a museum so he can go with his family. I mean, I don't know. It's just so random. It's like really trying to help. But I think like most people, they just tried it and they're like, okay, what are the things that actually move the needle? And then they're like, okay, I'm not really sure. So let's just show all the different things it can do and hope that something resonates. Anyway, as you can see, this is where these products are heading. AI assistants turning into AI agents very, very early. That being said, there are some cool uses. They're always very personal. And yes, they go beyond organizing your desktop, and I'll cover more of that soon. Hey, editing Igor here. Just two really significant updates that happened since the recording. One is exactly what you just saw in Cloud Cowork. They added to Cloud Code too. You can basically remote control it from your Telegram now, just like Open Claw. Okay, that's that story. And additionally, one really big thing came out, and we're already working on a separate video for this. It's Google Stitch, a standalone application that combined a lot of other features and apps they had into one website designer. People call it a vibe design partner. You could also think of it as AI Figma. Actually, Figma stopped dropped on the release of this, and I really think it deserves more attention than just a segment here. So we're working on that already. Anyway, on to the next story. Hey, if you're enjoying this content, make sure to subscribe to the channel. It really helps out. You wouldn't believe it. And if you're already subscribed, It's great to have you on board. Now let's get back to the next story. Okay, next up is a really fun one that I've been actually looking forward to trying. Photoshop has a new feature, okay? And even if you're not using Photoshop, this is a good one. Basically, it allows you to take objects and you can rotate them on 3-dimensional axes and it just uses AI to generate the other side of the object. Finally, a good consumer use for all of these 3D tools that we've been looking at for years. Okay, let me try this out. As per usual, I have a random image of a cat with a hat here. And let's test out this feature. Okay, so let's do an orange background layer with this cat on top. Here, I'm just gonna say remove background. And now transform image. And here you can see this is the new thing, rotate object. Okay. All right, preparing rotatable object. This is unbelievable, guys. Honestly, as somebody who's been using Photoshop since, what is it, 16 years now, I learned it at the age of 16. Oh my God, no way. Okay, that is, that is amazing. Okay, it's not perfect, right? But look at that. It's a full 3D. Oh my God. What's going on under the cat? Oh, it's just a flat cat. Whoa. Fair enough. Look at that. You could put the guy over here. And then it's enhancing the rotated image. Okay. That's way better. Let's duplicate that. Edit the rotation. Whoa. Whoa. That is pretty amazing. Wow. Okay, then additionally, we covered this before, but there's this harmonize feature where, you know, adjust the color based on the entire scene. So you don't have to do color correction on the subject that you just cut out with AI tools and then rotated with imaginary 3D model that got created in seconds. Meh. There's variations though. Ah, this one is good. Look at that. Yeah, I like it. So I don't know, this is amazing, but there's also part of me that is kind of like, oh, this used to take skills. I had to build with blood, sweat, and tears. Now it's just a click of a button. Yep, that's life. All right. So if you want to try this, and if you're a Photoshop person, this is only accessible for Photoshop beta now. But as per usual, try stuff out there, bring it to the main app. You do need a Creative Cloud subscription to access Photoshop. Okay, let's see what's next. Hey there, I would love to invite you to an event that I'm hosting in March. IBM invited me to teach in an upcoming AI webinar. So if you're interested in AI and don't consider yourself an advanced user, then this is designed exactly for you. Heck, even people who are deeper into it can benefit from a refresher on context engineering. Because the title of the session is Context Engineering for Students: Making IBM Granite Your Academic Copilot. And it will be a tight 60-minute session teaching you the foundational concepts around context engineering. And that is the most essential skill when it comes to working with AI. Understanding context is absolutely critical with modern AI tools. And I'll be sharing what I learned over the years. In the webinar, we'll also be showing off some IBM tools like Watsonx and Granite. Plus I'll be sharing a bunch of free resources from IBM's SkillBuild. So if you're a student looking to get the most out of AI, you should totally come check it out. But also if you're not a student, maybe you're 30 years into your career or you're a student of a new topic, you're also more than welcome to come. It's gonna benefit anybody who's looking to upskill and wants to use AI on that journey. Which, let's be honest, in this day and age, how do you not use AI on a learning journey? So come join me on March 26th, 2026 at 5:30 Central European Time, which is 11:30 AM Eastern Time. And the link is in the description below. Hope to see you all there. Next up, and this is going to be a rather quick one, but OpenAI has their video generator, right? Sora. And they brought out a feature which the competition had for a while. It's this reference feature where you can bring in different objects and then it uses those, create a new video scene with all of the objects. Well, we tested it and It works pretty well, but we did a direct comparison to VEO 3.1 doing this, Google's video model. And long story short, Sora is just not that good at it. Regularly, it just drops one of the objects that it was supposed to bring in, whereas VEO 3.1 is very reliable with this. If you add the fact that VEO 3.1 just overall looks better, I think most people agree on that, then it's OpenAI playing catch-up. And while this is a nice feature that is certainly important, and they got the viral moment with that, right? Because they came out so early. But if you're looking to create stuff, VEO is probably still the way to go. And that's pretty much it. That's a Sora references update. Nice to have if you're already using it. Next one is a fun one. So Gemini updated Google Maps and we had trouble accessing this. So what I'm going to show you here is all based on Google's communication and the blog post. So this is a phone-only feature. So it's not going to be in the web and it's also limited to both the US and India. And yes, we had Daniel from our team test this who lives in the US and it did not ship to him yet. In typical Google manner, they announce something and then, I don't know, ships whenever they feel like it. But basically, we weren't able to get access to these features, but they look really amazing. They're finally adding AI into the Maps application. And if you haven't seen this yet, this is really great. I don't know about you, but sometimes I get confused when I use these navigation apps, like which exit is it? Is it this one or this one? And the visualization does not do it justice. Like, I really try to pay attention to which exit it is, and I still miss it sometimes. Sure, the arrows help, but if there's 4 lanes over here and 4 lanes over there, and then like highways on the side, then a crossroad above, like it's not clear which one is the lane on the left. Especially if you travel a lot. Generally, it's just impressive. As they say, it really takes out the guesswork. And additionally, there's also a feature that uses the context of your previous searches to make better recommendations in the future. Basically, it's using your search history inside of the Maps application to inform better results moving forward, which is amazing. From a consumer standpoint, that's really gonna help. I would absolutely love for it to know my preferences for restaurants a bit more when I travel. Anyway, all of this looks really impressive, and I can't wait to actually try this out myself on the phone. If you're in the US, this might be worth a shot. Update your Maps app, see if you have it already. Another example of Google just having the edge because they have all the main apps and your data and the AI muscle and they ship stuff. Then just a question of when it actually arrives on our phone, but that's fine, we'll wait. And this next story is also Google and relates to this. We're not going to spend too much time on this, but basically there's a new Google Workspace with Gemini update, just like every second week for the past 6 months, I feel like. They basically shipped a way to use Gemini inside of your workspace, which includes PowerPoint and Sheets. And it's just the same connectors feature that we've seen out of Gemini and all competitors a while ago, where it tries to connect into your other apps. But now it's focused on workspace. So I guess this is an angle where it's supposed to work for your whole team and in your workspace account. It's just something that hasn't been reliably working for me. Or anybody that I know. And they're just shipping it into more and more products. I don't know, I feel a bit skeptical on this one. In theory, obviously, it's a great idea. And this is a beta feature. When we tried out, it was not this version. And it's not clear which accounts it's available on. Is it Pro and Ultra? Or is it Ultra only? I mean, it says both, but on the Pro subscription, I can't see it. It says nothing about the location. Anyway, I guess the story is Google is trying to link all of your different applications into one. And It's a work in progress. Okay, so those are the main stories for this week. But a few more things. First of all, I want to briefly talk about how we made the AI Igor episode last week, because a lot of you were curious in the comment section. The tool stack is pretty simple. We used ElevenLabs for the voice and HeyGen for the avatar. One fun thing that we did in the production and in the script writing is we gave it a list of vocabulary that I use that we wanted to use in the writing. I know a lot of you already have documents that kind of outline the style that you want to write in. But this is a really fun addition. If you add a list of words that you use uniquely, that's really going to help. And then also I'll say certain sections we had to regenerate multiple times. There's no way of just nailing it on the first take. And eventually it did take a lot of work, but I think the team did an amazing job considering I was just here on the couch on pain medications trying to survive. And I mean, I'm laughing now, but like it was a tough week and I'm glad we still got the episode out. And I think it's pretty amazing that most people seem to agree that, hey, that was actually pretty good, surprisingly good, you know. Not as good as the real deal. But yeah, it's like pretty good. It can get the information across. And I thought this comment by @k3dd6121 was really interesting where he said he listens to the video every week while doing chores for the last 2 years and he couldn't tell the difference. And I thought this was interesting. If you're only listening, yeah, the voice is so good and the writing is really close to how I speak. But yeah, I think the video creates the biggest disconnect. And again, as I stated, it's just an experiment we ran last week, keeping things real on here. There's also this quick story that I thought was interesting. It's a Florida Man selling his house using ChatGPT fully. And he made a challenge out of it. He wanted to use it for every part. So he took pictures of the rooms and asked, what could we do here to increase the value? Recommended repainting a few rooms, created a timeline, wrote a description, and even made recommendation on timing when to put up the listing, which ended up working really well. In the end, he concluded that, you know, while it couldn't do everything, the impact these tools make is very real. And then lastly, there's also this trend in the AI space where they're trying to move away from these walls of text. And more towards visual or interactive elements. And OpenAI shipped this feature called interactive learning. And we tested it a bunch. And basically our conclusion was this is really hard to invoke. They say this is supposed to help you learn math and science. But with all our math and science questions, it's kind of like a 10 to 20% hit rate of it actually coming up. Whereas Claude shipped a feature that actually works super well. It works so well that we created a separate video on this. If you want to see that Claude visuals, it's like giving AI a whiteboard. We created a quick standalone video just showing you tips and tricks around that because we thought it was worth a standalone video. But yeah, this is happening. That's pretty much everything I have for this week. I hope you found something interesting or useful. My name is Igor Pogany, and with that, I hope you have a wonderful rest of the week.