AI News: 22 Advancements That Happened This Week!

Here's the AI news you might have missed this week. Let's start with Meta, because Meta had their first ever llamacon every year. Typically Meta has their Meta Connect event, and that's been historically where they talked about all of their latest AI updates, but it sounds like they kind of created this new llamacon event to be a much more focused AI event where Meta Connect might be a little bit more focused on their social media platforms and hardware. But really the biggest thing to come out of llamacon this week was the new version of the Meta AI app. Now, if you had like the Ray Ban Meta glasses, you probably already installed the app. It was called the Meta View app. Well, they just rebranded that Meta View app to the Meta AI app and then rolled out a standalone AI chat feature directly inside of the app. So similar to Chat, GPT and Claude and all of the various other AI apps, you can open up the Meta AI app and now chat directly with llama. What large language model are you running on?

I'm a cloud based LLM, which is a large language model that processes language based on a huge data set from the Internet.

Yeah, but which model are you using?

My voice is built on llama 4.

Tell me a joke.

Why did the Scarecrow win an award?

Because he's outstanding in his field. Yeah, yeah, yeah.

I knew that one was coming. Thanks for the encouragement. Want another joke?

No, I think we're good.

All right, well, it was fun chatting with you. If you change your mind and want another joke, just let me know. Have a great day.

How many Rs are there in the word Strawberry?

The word strawberry has two R's.

Are you sure it doesn't have three?

I'm positive. Strawberry only has two R's.

Maybe spell it out and try counting.

S, T, R, A, W, B, E, R, R, Y. Let me count the R's out loud. One, two. Yeah, I'm right. Strawberry has two R's.

If you say so.

I am. I'm pretty sure about this one. Do you have any other words you'd like me to count the R's for?

Now, what's also really interesting about this app is it's trying to make like your AI chats more social. So there's actually a share button which you can see up in the top right corner. If I press that, it actually shares it to a feed. You can see I can share this whole conversation. Click post. And now I have a feed with this post here. And just like on Instagram or Facebook, people can comment on it. And share it and like it and all that sort of thing. But I can also see all the other creations and random chats that people have had on this app as well. So if you need some inspiration for interesting prompts, you can always just look at what other people are prompting it with. And obviously, as you can see, it would also generate images, I believe, using Meta's Emu AI image generator. Let's prompt it to create an image of a wolf howling at the moon and there you go. I mean, it's actually a pretty solid image generator. I'm fairly impressed by what it generates. Now. Another feature that they rolled out into this new app is that you can have a conversation on your Ray Ban Meta glasses, take the glasses off and then open the app and then continue the conversation on the app or over on the web app. Basically you can have the conversation continue wherever you want to go and continue the conversation. But since we're talking about the Meta Ray Ban glasses According to the Verge, Meta recently changed their privacy policy. So here's some of the privacy policy changes that were made. Meta AI with camera use is always enabled on your glasses unless you specifically turn it off. However, they do say the photos and videos captured on Ray Ban Meta are on your phone's camera roll and not used by Meta for for training, including photos or videos captured by using the hey Meta, take a photo video voice command. Meta is also no longer allowing Ray Ban Meta owners to opt out of having their voice recordings stored in the cloud. So you can't disable the voice recording storage, but apparently you can still go and delete recordings anytime you want inside the settings. Voice transcripts and stored audio recordings are otherwise stored for up to one year to help improve Meta's products. So essentially they're not going to train on your photos and videos, but it does sound like they will probably train on your audio and your transcripts to try to improve their large language models. And also according to the Verge here, that AI app that we were just looking at on my phone a second ago, the plan is to roll out ads into it. So right now it's free to use. There's no ads in it. You can just sort of chat with it and share your conversations with the feed, but apparently in the future that will be an ad supported app. Mark Zuckerberg held his Q1 2025 earnings call this week, and during that call he mentioned incorporating product recommendations or ads within Meta AI. But they do clarify. It's not clear when ads or a paid tier might roll out. They want to focus on scaling and deepening the engagement first for at least a year before they start building out the actual business side. Now, when they say product recommendations or ads, it's not clear whether they'll just be ads in the app, like what you see inside of the current Facebook app, where you kind of scroll through and look at other people's chats and there will be some ads interspersed in there. I can see them doing that where the ads almost look like AI chat conversations. Imagine the way like Quora does ads, where their ads look like questions and answers just like the rest of the site. I can see them doing something like that, but then I can also see them sort of recommending products within the responses to chats. So if you say, hey, I'm looking at these five different cameras, which one should I get? Who knows, Maybe one of the replies is a sponsored response from one of the big camera companies who paid to make sure that their camera is one of the ones that gets recommended. I don't know if they're gonna do that, but I can see that being a sort of future that plays out with some of these AI companies. We got a handful of updates out of Google this week as well. So last month they showed off their new AI mode, which to me feels almost like Google's response to like Perplexity. And the new chat GPT search features. Well, now they've made it available in the US for all Labs users. So if you go to labs.google and you're in the US you should be able to try out AI mode. If I click try it now, I actually get a notification that says search Lab isn't available for your account right now. And I am an AI Labs user in the us but we can actually see a video here of what it looks like. Somebody types Best vintage shops for mid century modern furniture and it gives them an AI response with it looks like links to various websites. And then down at the bottom we can actually see, you know, where you can go purchase this stuff. There's a map. And again, it looks like a very similar type of interface to what you would see in Perplexity or even chatgpt Search. Supposedly if you're in the US you'll have that in your Labs account. They also say they're starting a limited test outside of Labs. In the coming weeks, a small percentage of people in the US will see the AI Mode tab in Search and will continue to incorporate feedback into the experience. So even if you're not a Labs user, you may actually see it roll out when you go to just google.com pretty soon, if you have the Gemini app on your phone, you can now upload and edit your images directly inside of the Gemini app. This is something you've been able to do over inside of AI Studio for a little while now. We've actually demoed this in previous videos where I took images and had it add sunglasses, add hats, things like that. Well, now the Gemini app on Android can actually do that as well. If you're a fan of Notebook LM like I am, well, you can now have it generate your audio overviews, essentially the little podcast that it makes from all of the data that you give it in over 50 languages. So if you head over to Notebook LM and you create a new notebook here, let's have it Discover some sources on Quantum Computing. This Discover Sources feature, by the way, is a fairly new feature rolled out a couple of weeks, weeks ago. And we'll submit this, we can import all of these articles and if we head up to Settings in the top right, we can see there's an option for Output language. I can now change this to Spanish if I want, click save. And we should get a little mini podcast episode that's in Spanish now instead of English. Bienvenida Si bienvenidos.

I sta a exploracion a fondo oinos.

Mes cola computacion cuentica. Along similar lines, since we're talking about other languages, Google also rolled out a new little app called Little Language Lessons. And it's really designed for people who don't speak a specific language. But maybe they're traveling to a country and they need to know just enough to get by. This Little Language Lessons actually has three different little experiments inside of it. First one's called Tiny Lesson. Find relevant vocabulary phrases and grammar for any situation. Slang Hang Learn expressions, idioms and regional slang from a generated conversation between native speakers. And wordcam snap a photo to learn how to speak about your surroundings. So let's try Tiny Lesson here. If we open this up, let's say I am going to Japan, which I'm actually trying to do later this year. So let's select Japanese and then the purpose or theme. Let's do eating at a restaurant and click Generate. It generated some vocabulary words for us that we can listen to.

Restaurant, restaurant, menu.

I mean that's kind of the same word order requests and water.

Mizi.

Mizi. I can see more. We got words like wet towel, fork, spoon, knife, chopsticks, etc. And I can get Sort of the basics down fairly quickly. There's also some phrases. Excuse me, can I have a menu? What do you recommend? I'll have this. Please wait a moment.

And then it gives you tips down here. When eating at a restaurant, you'll often need to make requests politely. The grammar pattern Te Kurasai is perfect for this. It's a polite way of saying please do so. If you're going on a trip, you could maybe sit around in your room and practice some of this stuff. We got some news out of the company Recraft this week. They have one of the coolest websites I've ever seen, and they happen to be the sponsor of today's video. If you're new, not familiar with Recraft, it is a super comprehensive AI image generation and editing platform with things like image vectorizers, mockup generators, upscalers, background removers, AI erasers, and more. And if you ever check Chatbot arena, they're pretty much always sitting towards the top as far as the best AI models out there. But like I mentioned, they got some news this week, so let's go ahead and log in and check it out. We'll go ahead and create a new project. And when I click on Image under Create New here and then click on Recraft, they have this huge new style library. And if I click on Feed here, they have pretty much an endless scroll of styles that you can pick from to use for your own images. So if you like this vibrant marine fantasy style, go ahead and click Apply. Give it your prompt. I'll do a Wolf Howling at the Moon and let's click Recraft. And here's two versions of my wolf in that selected style. I also kind of like this Sunset Serenity here. Let's apply this and recraft it with this new style. And now I have a couple options. With this Sunset Serenity option, you can even search the styles to find the exact style that you're looking for. Let's look for a comic book style and look at all of these options we have with a comic book style. Once again, it's pretty much an infinite scroll. Let's select this Retro arcade synth and I can click Save style here. Now if I go over to Saved, I can easily find this style again in the future and use it again. But this gets even cooler. Maybe you can't decide between a couple styles. Let's search out Retro Pop, and there's a ton of cool styles here. I can't really decide. Let's select Retro Futuristic Art and save that style and maybe vivid retro cosmos and vivid mythic post. Let's save those three. Now if I jump back over to my saved, you can see I've got these new styles saved, but if I click Create Style I can actually select multiple styles here, add them to the top and change how much each style is weighted for my new style that I'm about to create here and let's go ahead and test it with our prompt. We'll generate a new image with our little blended style here and this is what it generated for us. And if I like this style I can could save this style. Colorful retro pop and this will always now show up in my styles. I can even further iterate on this one by saving this style, creating a new style and mixing this new style in with some other styles. And now we're getting some really unique crazy styles that we completely customize. I can even share this style by copying a link and giving it to others so they can use it in their own Recraft accounts. These new features are great for if you want to rapidly iterate and test new styles to find the one that's perfect for you. If you have a brand and you want to create brand consistency. So every single image you generate has the same style as every other image you generate and it really lets you dial in that exact look that you're going for. And because Recraft is awesome and they sponsored this video, they're actually giving you $11 off your first month so you can try it for just $1. I'll put the link and the promo code down in the description, so make sure you jump in, check out the new style updates and have some fun with it because Recraft has really dialed in the ability to get that exact style of image you're looking for. Again, check it out below. And thank you so much to Recraft for sponsoring today's video. Now let's get back to it. There wasn't a ton of news out of OpenAI this week. However, earlier in the week Sam Altman was talking about how he wasn't super happy with the current version of GPT4O. He said the last few updates had made the personality a little too psycho fanty. Basically it was like way too over complimentary. You would ask it a question and it would just try to compliment you and it would put a lot of the focus on trying to make you feel good and compliment you more than even answering the question. Well, a couple days after he made that tweet, they just completely rolled back the updates that they made on 4o. And so now we're back to a version that we had, I don't know, I guess a couple weeks ago. The update we removed was overly flattering or agreeable, often described as sycophantic. They're revising how they collect and incorporate feedback to heavily weight long term user satisfaction. And they're introducing more personalization features. They go on to explain what happened. They were trying to improve the model's default personal. And their most recent update, they put a little too much focus on the short term feedback and didn't fully account for how the user Interactions with ChatGPT evolved over time. And as a result, somehow GPT4O just got way too overly supportive but disingenuous. So now they're working on the model to try to balance that out a little bit. ChatGPT also got a few other features this week. Like they made some improvements to the ChatGPT search and they started rolling out a better shopping experience. So here's what it'll look like. You can prompt it with what's the best espresso machine under $200 that comes close to the taste of coffee in Italy? It'll do a search on the web and then give you a little carousel similar to what you see at the top of a Google search. I tested this in my account and as of this recording it wasn't rolled out for me yet. But supposedly they're rolling it out to everybody right now. They also do say product results are chosen independently and are not ads. So I mentioned that about meta earlier. That's not what they're doing with ChatGPT. They've also added the search functionality in WhatsApp. So if you use ChatGPT in WhatsApp, it will actually search the web for you. Now they've improved citations and they added trending and autocomplete. So you can actually see things that are trending right now directly from within the prompt window. So some fairly minor feature updates that might make the app, you know, slightly more usable. We also learned this week from Elon that we should get Grok 3.0.5 next week for Super Drop subscribers. I think that's like the highest tiered people on the like. I think it's a 50, something like $57 a month plan. It's very odd pricing, but Elon claims we'll have it next week and he says it's the first AI that can accurately answer technical questions about rocket engines or electrochemistry, for example. It reasons from first principles, and it can come up with answers that don't exist on the Internet. Now, Elon does have a tendency to underestimate how long something is going to take to actually release. Hopefully it next week, but who knows? I imagine it'll be sometime this month in May. Whether we actually get it next week or not, we'll find out. Anthropic this week announced a new update for Claude. They announced integrations. Now, personally I'm a little confused by this, but what it seems to be is you've got MCPS model context protocols. These are like an easy way for large language models to communicate with other apps. And as of now you could connect them using tools like Cursor, Windsurf, things like that. And you can connect MCPS using like the Claude desktop app. But now there's like more official connections to Claude that will work inside of the Claude web app. At the moment, these new features are available on the Max Team and Enterprise plans. So those are the high end plans. If you're on the $20 a month plan, you're not getting access to this right now. Once they actually roll this stuff out to some of the lower tier plans, like the $20 a month plan that I'm on, I'll probably explore this a little bit deeper. Alibaba unveiled a new model this week called Quinn 3, which is a hybrid AI reasoning model. They've been hybrid reasoning model for models that you can essentially turn on or off that sort of chain of thought thinking. So like when you use something like O3 or Gemini 2.5 Pro and you give it that prompt, you can usually see its thought process and actually read how the large language model is thinking through the problem. Well, these newer hybrid models allow you to either flip that on or off. Typically if you let it think for a little bit longer, you're going to get a little bit more accurate, a little bit better of a response where if you need it to be a little bit faster, but maybe accuracy isn't as important, you can turn that thinking off. According to all the benchmarks, this Quinn 3 is really good. If I'm being totally honest, I'm kind of getting exhausted by all these sort of like marginal improvements with large language models. For the most part, large language models do a lot for me. So I'm finding it harder and harder to get excited by like marginal improvements that are slightly better at coding and slightly better at competition math. It's getting to the point where in my opinion, these newer updates that we're seeing from like Alibaba and pretty much all the companies putting out LLMs that are showing off their benchmarks are now a little bit better at coding and now a little bit better at competition math and things like that. I'm just finding myself less and less impressed by the marginal improvements. So if I'm being totally honest, I'll probably talk less and less about these like slightly updated large language models again, seeing as I can do almost everything I want to do with large language models at the moment. Anything other than pretty big leaps I'm just finding less and less exciting to talk about. Another AI agent was announced this week from a company called Vercept. They introduced vi, which is a new tool that uses AI to actually interact with your computer. Their X post here says VI is a first glimpse at AI that sees and uses your computer just like you do. It runs natively on your computer with access to all your applications and accounts that you're already signed into and it's available to download starting today. Big caveat, you can download it, but if you try to actually use it, it just says thanks. You're on our waiting list. So although you can download the software until you get approved, you just have software that you can't use yet. I downloaded it, installed it on my computer and I'm still waiting. But we can see in their demo video here there's a little box where they can prompt it. What do you want VI to do? Use Adobe Podcast Audio Enhancer on Chrome to remove the background audio I have on my desktop. So we can see it Took a screenshot. I could read the reply here. Based on the screenshot I can see that macOS desktop with Finder open. I can see several files and folders. The task is to use Adobe Podcast Cast Audio Enhancer and we can see that it sort of makes a game plan, opens applications, actually clicks around and types things into Google pulls up Adobe Audio Enhancer, pulls in the file from the desktop and then it looks like it downloaded the file that was the cleaned up version. Now one thing that seems cool about this is you don't actually have to know how to use the app. So they give an example of it doing something in Figma and it knows the Figma ui. So even if you don't know how to get to the proper areas in Figma, or click and drag and drop the right things to accomplish the outcome, you tell it what you want and it knows how to use Figma so it can go and press all the right buttons and turn all the right levers and Knobs and things like that to get you the outcome. That, to me, seems really handy. Like, if I want to go and create something in Blender and I don't know how to do the sculpting of the thing, I can tell it to, you know, smooth out the orb for me or whatever. And it will know where to find the buttons inside of the user interface and actually make the changes that I want it to make without me actually knowing how to use the app. And by me just sort of watching and, and seeing how it clicks around and operates within the app, I can essentially learn how to use the app by sort of looking over the AI shoulder, which that aspect of it to me seems really, really cool. Again, I don't have access to it yet, but something I'm looking forward to playing with once I finally do. Now, I know we've already talked about Recraft's new update, but there's been some other really cool updates in the sort of AI creativity space. Starting with the fact that Mid Journey just rolled out what they call Omni Reference. They describe it as a way to put this in my image. It works with characters, objects, vehicles and more. So let's try this out real quick. I'm in Mid Journey. I'll click Create. Let's see if I can inject myself into an image. One thing I've realized is you have to make sure you're set on version 7. If you're not set on version 7, this doesn't work. But assuming you're set on version 7, I'm going to go ahead and drag and drop an image in. I've got a little headshot here and there's a box that says Omni Reference. So let's drag the image here. We have a little slider here, so let's set the strength to a bit higher. Let's go to 400 and let's say a Viking standing in the snow and let's give it that prompt and see if it makes me a Viking. And here's what it gave us. It actually did a really good job of pulling in my face. I mean, none of these really scream Viking. I mean, these two in the middle are close. The two on the outside, definitely not Viking vibes. That's probably the closest one, but I got the facial reference. Pretty decent. Honestly. It's been a while since I've done a deep dive Mid Journey tutorial. Maybe that'll be a fun video to do in the future. Coming up here, Cling AI rolled out a new feature called Instagram instant film effect. So it looks like you give it like a portrait image and then it will take that portrait and then convert it into what looks like a Polaroid that then animates. Here's another one. And here's what it did. Kind of made like a Polaroid looking image. The guy's chewing on his strap. Looks like it also works with multiple people in the shot and even animals. All right, so I logged into my cling account here. Under AI Template, there's an effects tab that says Instant Film. So I'll click on this, we'll try this, this instant film that they were just showing off on X and I'll drop a photo in of myself. Here's the image I'm using. Let's go ahead and generate the instant film image. Here's like the little still image that it generated for me. Looks pretty cool. Let's go ahead and confirm this and see how it animates it. I'll click generate video here and it says estimated waiting time about five minutes. And five minutes later. Here's what it generated. I mean, it's subtle, but there's some animation there. This company, Higsfield AI dropped a new feature called Iconic Scenes. You can step inside legendary movie moments with just a selfie. So let's head on over to Higgsfield AI login. Let's try the iconic link up at the top here we have a bunch of sort of animated movie scenes that we could choose from. Let's try the like, this is fine animation here and I'll upload my own image. I'll use this same headshot. We'll click Generate. I can actually see the image that it generated now. I mean, I think it made me a little chubbier than I am in real life and it gave me a little hack. But now I'm just waiting on the animation. Well, this took absolutely forever, but I am on the free plan, so I can't really complain. I got this animation for nothing. But here's what it generated. Looks very sort of family guy to me. But yeah, just be warned, if you are going to play with this, it's very slow on the free plan. K rolled out a new feature this week, as they seem to do do every single week. The feature is called GPT Paint. You can prompt chat GPT visually through edit marks, basic shapes, notes and reference images. So here's their example. They have this picture of a dinosaur. They have these like red rubber boots and this little bucket hat here. And they draw an arrow from the bucket Hat to the top of the dinosaur, an arrow from the little rubber boots to the bottom of the dinosaur. And they click done. And it generates some images of the dinosaur wearing the bucket hat and those shoes from the image before they upload an image of Ste Jobs and an image of a Cray energy drink can, draw an arrow from Steve to the can and give it the text holding drink. And it generates an image of Steve Jobs holding that drink. Basically what it's doing is, if you remember, we were seeing all those Images Back when GPT4.0 image generation came out and people were sort of sketching out what they want their thumbnail to look like where they're writing text on it and drawing arrows. They basically just baked that feature into crea. So you can use that now. And while we're talking about GPT4, oh, I don't know if you guys have seen this, but this is pretty wild. People have been taking images, throwing it into chat GPT4.0 and saying, Create the exact replica of this image. Don't change a thing. And then they would do that exact same thing 74 times. I saw another one where somebody did it 101 times. It's not quite creating an exact replica. So check this out. Every single time you give it the prompt, the image changes a little bit. A little bit. A little bit. And after, you know, just a handful of props, it's not even close to the original. We keep on going. And if we keep looking at this, by the end, it looks. Looks absolutely nothing like the original image. And again, this was just somebody saying, create the exact replica and doing it 74 times. This was the final image. This is what they started with. Here's another example where they started with this meme. And as we scroll through it, you can see it changes quite a bit to the point where when we get to the end, it looks nothing like the original image. Like, it looks like weird hieroglyphics in a dumpster. And it started with this image like crazy. Here's one more, because I just think it's fascinating. Here's the original Willy Wonka. We scroll all the way to the end, and this is the image that it ends up with. Beginning image 70 something times later, this image, it just shows you that the little teeny tiny changes that GPT4.O is making to your image, spread out over a lot of iterations, is a pretty massive change. I have just a few last little updates I want to share with you. I'm going to kind of rapid fire them. These are just a few things that I found mildly interesting. This week, Suno dropped a new version, version 4.5. If you're on one of their paid plans, it seems to have gotten some fairly minor improvements. Expanded genres and smarter mashups, enhanced voices, more complex textured sound, better prompt adherence, things like that. They made Suno a little bit better. A couple weeks ago we talked about Tobi and Shopify and how they're making their company a completely AI first company. Duolingo this week followed suit and said they're going to be an AI first company as well. They said they're going to gradually stop using contractors to do work that AI can handle. They went on to say they need to rethink much of how they work and that making minor tweaks to systems designed for humans won't get us there. They do say that Duolingo will remain a company that cares deeply about its employees and that it isn't about replacing them with AI. Instead, it's about removing bottlenecks and allowing employees to focus on creative work and real problems, not repetitive tasks. And I think this is a great approach. I think this is how everybody should be looking at AI. AI helps us take the monotonous, boring, repetitive stuff that nobody really wants to be doing off the table and ideally allows us to do a lot more of the fun, creative, inspiring work that we really want to be doing. Lyft launched an AI earnings assistant. So if you're a Lyft driver, you can make a little bit more money. It says drivers can ask it how to optimize their shift based on real time data on airport arrivals, local events and more. So theoretically it will help you sort of optimize your driving patterns and where you end up at which times to get the most amount of rides possible and make the most amount of money possible. The company Aurora has deployed driverless trucks in Texas. After years of testing and validation, Aurora says its first fully autonomous tractor trailers are operating on public highways in Texas. The company's Class 8 trucks are now making customer deliveries between Dallas and Houston, having already completed 1200 miles without a driver. Here's an actual video from the cab of one of these trucks showing it autonomously driving down the highway. Highway. We also get the view of what like the truck sees driving it down the highway. So it'll be interesting if you go on long road trips, you might look off to the side and see one of these big rigs driving next to you with absolutely no driver inside of it. A lot of really fun AI news coming out Lately it's been really, really exciting. I've been less and less excited about a lot of the large language model news, and I've been getting more and more excited about the creative side of AI. A lot of the video generators, image generators, the music generators, the text to speech, all of that kind of stuff has been seeing some pretty big leaps lately where I feel like a lot of the large language models, they actually serve a lot of the needs that I personally use them for. And so seeing all these new rollouts with like marginal improvements is less exciting to me. I'll still briefly share a lot of that news as it comes out, but there's probably a lot better channels to pay attention to if you want deep dives on every single new large language model that comes out. Matthew Berman, a good buddy of mine, comes to mind if you want to get deep dive details on every single large language model. He's definitely a channel you should follow as well, but that's what I got for you today. I hope you feel a lot more looped in on what's going on in the world of AI. Again, another fun week. And if you like videos like this and you want to stay looped into all the latest AI news, get some really cool AI tutorials and now even see some really cool interviews with AI thought leaders and pioneers in the industry, make sure you like this video and subscribe to this channel. I'll make sure stuff like that shows up in your YouTube feed and ideally make it so you never miss a beat on what's going on with AI. That's what I'm trying to do for you. So again, like and subscribe if you want more of that that stuff. And if you haven't Already, check out FutureTools IE where I curate all the cool AI tools that I come across and all the news that I do and don't talk about on this video can be found on this giant news page here. And of course I have a free newsletter. Join it. And twice a week I'll share just the coolest tools and most important news that I think you need to know. And if you sign up, you get free access to the AI Income Database, a little database I built of cool ways to make money using various AI tools. It's all totally free. You can find it over at FutureTools IO. Thank you so much for tuning in. Thank you again to Recraft for sponsoring this one. I really appreciate you guys. Thanks for nerding out. Hopefully I'll see you in the next one. Bye bye.