Copilot is Finally Useful, New Claude for Excel & More AI Use Cases

Welcome to yet another week in AI. This time around, as you can see, I'm on the go. I'm in Arizona in the USA right now. Slightly jetlagged, not going to lie. But there's a bunch of stories for us to cover and try here, including a new version of Microsoft Copilot, a streaming AI video generator. That's an interesting one. And then new updates from Adobe, Claude, Chat GPT, and more. In this week's episode of AI News, you can use the show that pulls together all the AI releases of this week. Filters for the ones that you can use. Well, mostly this week we had some limitations and the ones that matter and then I get to present it all back to you. Let's

begin. Okay, so here we have a brand new version of the Copilot app released. And I know what you might be saying. — Hm. Copilot, the app that generally considered people only use if they have to because their company has a license and they don't allow any other AI. — But we tried this out and we were pleasantly surprised. There's a bunch of new features in here and it's really starting to separate itself from the pack in a little ways for now, but this is becoming a really robust product. They added essential features on other platforms like memory that you might already know from Chachi or Claude connectors, a first version of actions. And then this one called groups, which I found particularly interesting. It's a conversation with an AI that is multiplayer. So multiple people can join. And when I type in here and get an answer, Daniel sees the same thing, but all three of us are in this chat. And you can invite more people. Honestly, I'm not exactly sure what the point here is. I guess you could collaborate on projects. It kind of like seeds this chat and then other people could work with it. It's really interesting. I haven't seen this in any other one of the main applications. Kind of nice, honestly. Multiplayer chats. They don't call it this way, but I call it multiplayer chats. And then there's also new image generation and remixing features. So overall, I love to see them go in a new direction, take the product and seeing something I would call experimentation from Microsoft AI, which is good. I kind of want to see more of that. And I think amongst all the other features that we'll talk about here in a second, the one that stands out is they added a character into here called Mo, which is sort of like Clippy from back in the day, if you remember. Apparently, there's even a Easter egg where Mo turns into Clippy when you repeatably tap Mo. [snorts] I like that. Overall, I think this is a great idea, giving your AI assistant a personality. It's a different approach from what OpenAI, Claude, or Gemini or X or Meta have right now. All right, so if

you build, run web apps, APIs, dashboards, or similar, this next one is for you. Maybe you started recently and you're already outgrowing a hobby hosting and need something simple, scalable, and predictable. Well, in that case, you need to check out Savala, the sponsor of today's video. Savala is an all-in-one platform where you deploy your application, your database, and your static site all in one piece. And that's a wonderful thing about this. There's no trickery, no surprise bills. If you try to build and host things before, you might have woken up with a $400 headache because your app scaled in the middle of the night. And many other platforms love these limits like seeds, tiers, build cues, and then just later on you get the bill. That's not going to happen with Zavala because they do usage based pricing and the traffic in between your components is free. So when your app talks to your database or your storage, the meter isn't secretly running. That alone can save real money and more importantly, it's predictable. And I just want to underline how good their team facing products are cuz you get some unexpected perks here. Unlimited collaborators, no seat licenses, no fixed plans, unlimited parallel builds, uncommon in this day and age, actually. You just bring the whole team and don't have to worry about the math of that. Plus, you get enterprisegrade security and real human support from developers. So, if you're looking for state-of-the-art features and real transparency in pricing, check out Savala today. You can get started with a $50 credit that is auto applied to all new accounts. A big thank you to Savala for sponsoring this video and making these productions possible. And now, let's get to the next piece of AI news that you can use. Okay, so this

story is a really good one. It's called Google's Pomelli and this is yet another one of their labs experiments where they built various AI applications and I think this one turned out particularly well. So in their words, you can create onbrand marketing content for your business. The way it works is quite simple. You begin by scanning your website and it creates a business DNA which you can see me doing right here. And just a few minutes later, you get this entire business DNA profile with your fonts, your colors, etc. So in my case, it did really well except of the logo. But you could always replace that, which doesn't seem to work perfectly. So then you could also generate. It feels like right now they're experiencing traffic issues. So, for example, if I start the process and use the website of the Coca-Cola company just to test this. And yeah, also when I tried the Coca-Cola company, it doesn't manage to actually pull the images, which is a problem here because that's kind of the point because let me show you what happens when you go on. During some of our testing, this worked really well, but by now it got so much traffic that this doesn't seem to be the case anymore. So if I pick one of these campaigns, well then it's supposed to take all of those assets and remix them into a custom marketing campaign. But as it has problem with those images, doesn't seem to work so well right now. You're supposed to be seeing is something like this. Various marketing images, social media posts with a variation of ad copy, all pre-made. And I wish I could show you, but we tried yesterday, we tried today. Just seems to be overwhelmed right now. A shame really, but not much I can do about that. Maybe we could circle back next week. And for now, let's see what's

next. Next up, we have Chatd Pulse rolling out to the web. So, if you remember, this was their mobile feature where it proactively suggests stories and things you could be doing every day. And if I go into the web version and I go to settings, personalization, and I have all the memories features on. And then down here, you can activate pulse. And if you go to this new tab here on the left, which says pulse, well, then I can start using this. So for yesterday, like 10 minutes ago, October 30th, kind of summarizes the theme of the week, which is accurate actually. And it shows me different use cases and features. And then it gives me options for future updates, like can set some goal that you might have been working on recently. Or let's just do one of these presets here. 1 minute daily momentum reset. I like the sound of that. Or this little pre-generated article on why context switching drains mental energy. That's a relevant one this week. So, as you can see, this is just a different approach to chat GPT and I think one that we'll see more and more of in the future. As I mentioned before, practically tells you what you could be doing rather than you having to take action and to prompt it. I think this is really going to unfold its full potential once it has access to more context on you. That's what they're trying to do with their browser play. Definitely an interesting thing that anybody on a pro account can now try. Yes, that's a $200 one. Other paid accounts have it on the mobile app, not

on the web one yet. Okay, for the next story, we have Adobe's Max happening this week. And obviously, they're shipping a bunch of new features, including AI features. And I want to highlight this one, which is new Adobe Firefly releases, which yes, there's new models, but they also remodeled their studio interface, which is also built into applications like Photoshop. So, as I showed you in previous weeks, now you can also select some other third-party models inside apps like Photoshop. So you could use Nano Banana within it, but also in this web interface, you can see that now it can do a whole lot more than just image generation. If I go to video, I get all of these choices. I don't just get Firefly video, I get all of these other choices like VO3. 1 even gives me things like the first frame feature. And it might be a bit tucked away, but you can also do things like speech. So up here you only see image and video, but if you go here, you can see that I can enhance speech, do text, sound effects, I can generate speech, I can even create these boards, a bunch of quick action presets and community inspiration down here, all in one place. So let's just try and generate speech. Just use defaults. — Watching news you can use. — And again, that's decent, but I could switch this to the 11 apps models, which are state-of-the-art. — You are watching news you can use. — Much better. So generally I really like this idea of a studio for all these things and especially these quick actions. Have a look at this quickly caption videos. They even have their own clip maker now which I suppose is going to compete with all the other clip makers that kind of don't really work. I mean if you give it podcast I guess one out of 10 of them is okay. But anything that's properly edited, none of these clip makers really make the cut. And honestly the thing that just has worked the best for us in video production is the enhanced speech ring. This actually came out 2 years ago. And if you're not aware, this is one of the best AI creative tools. It's just magical when it comes to voice. — Adjusting the strength slider to see how it does at uh canceling out the background noise and enhancing my voice. One more example before I go. This is an elevator lobby uh also here at the Adobe office. Very echoey, unbelievably good tool. It is entirely copyright free. And this is their big claim also with their image and video models that all of their models have only been trained on copyright free materials. And if you're using this, you can have 110% confidence in the fact that you're not breaching any copyright laws. With some recent court decisions in the US stating that training upon public data isn't copyright infringement, this might not be as relevant as it might have been a year ago. Yet still, it's sort of this gray area and some of those court decisions aren't the final verdict for all of humanity forever. And if you're taking a lot of responsibility for big campaigns, these models might be a way to improve your sleep at night. and then progressing on this idea of a studio with all the apps in one place. It's good. I love easy to use interfaces like this for creative stuff. This is quickly becoming a very valuable hub where you get a lot of the best stuff in one place rather than having to have 10 different separate tools, websites or platforms.

Okay, not bad. There was one more story that I was itching to try this week, but I couldn't get access and that's Claude for Excel. It's an extension that is purposely built for Microsoft Excel and to optimize to work with it. Really multitab worksheets, advanced formulas, all that. And from what I've heard so far, this is supposed to be by far the best way to work with Excel and AI. Better than all the native solutions, other apps. But unfortunately, this is behind the weight list for now. And as of now, they only rule it out to the first thousand users, which this time around is not me with my account. So, we'll definitely circle back to this once I get access to this. But basically, it seems like they're rolling out something like their browser extension, but for specialized apps, which makes a lot of sense, and I think we'll see more of in the future. I just can't demo this thing for you right now cuz I don't have it. All right, let's

move on. The next one is going to be very interesting. It's a new Wikipedia generated by Gro Groipedia. Basically the claim here is that this is an independent uneditable source of truth as opposed to Wikipedia where many individuals have the rights to edit the articles and apparently I haven't been following this closely a lot of the articles have become politicized and now represent certain worldviews from what I've read on it generally more on the liberal side. I'm not here to discuss politics. I'm here to look at AI innovation and using AI to regenerate an entire Wikipedia certainly categorizes as such. So, what I kind of just want to practically do here is just pull up an article and compare. How about the article on the United States? I mean, that seems like a good place to start. Okay, so first of all, you can see that Groipedia doesn't have any images yet, I suppose. Just a big wall of text Wikipedia. It's kind of mixed. And then if I just look at the first paragraph here, it's essentially the same information in a different order. So, as a lot of the argumentation why they created Wikipedia is political. Let's go into government and politics. Let me actually do something super random that I just thought of, which is taking the words Democratic Party here, but I'm just going to search for it. In Wikipedia, it occurs two times here and here. In Groipedia, one time and then Republican Party, same thing also one time. Okay, that didn't yield any results. Maybe Ukraine. Okay, seven times. Two of those are within the articles and five within references. And within Groipedia one time, okay, that's the difference. Let's look at what's written here. Actually, Gipedia mentions how much aid was provided before October 2024, whereas this article gives way less detail and no figures and it even mentions relations with China as a part of this conflict. Whereas Wikipedia, I guess, sort of avoids a topic. I mean, look, I literally took a random sample, took the first thing that I thought of and just looked for two keywords. So, this doesn't mean much, but I kind of just wanted to explore this idea of there being a source of truth for ideas. And I mean, generally speaking, independent of both Wikipedia and Groedia, I just like the idea of there being a neutral source of truth that is uneditable. I mean, for example, I hate the idea of certain studios or copyright owners of books going back and editing original works just because the political situation might be shifting. I personally think that stuff is absolutely unacceptable. And I don't know if this is the implementation we need, but the idea of there being an independent source is right. And in this little example, yeah, in this little paragraph, I would prefer Groipedia over what Wikipedia tells me here. But then I guess you're giving up all of the images, which might not matter because this could be used as a sort of true for LLMs rather than newer models having to rely upon Wikipedia or again literally some individuals can go in change facts, therefore just rewrite history. One more thing I want to mention here is that Wikipedia has over 7 million English language articles whereas Groedia right now is at 900,000 and all of them are fact checked by XAI's chatbot Grock. Anyway, just wanted to cover that schemes relevant especially in times where there's so many versions of the truth. It just entirely depends on who you ask. I like this idea of having a central source. I expect there to be more versions than just this one. This is just the first we got. Okay, let's

move on. Oh, and this next one should be a lot of fun. It's called Orisa 2 and it's a instant interactive AI video model meaning this streams AI video and as I prompt it should be changing. So let's just start with a cat with a hat in space flying through space with a crew of space chihuahuas. Okay, that's sort of Yep. Okay. Create a black hole. A cat turned into a black hole with tacos coming out of it. Okay. And croissants filled with green goo being sucked into the black hole. Hey guys, now you get a little insight into what's actually going on inside of my head when I'm not recording this video. This is just stream of consciousness prompting. Oh my god, this one is weird as hell. It was pretty good though. What is this robot though? Zooming into Central Asia. Can I do a transition like that? This is a lot of fun. Good thing they put me on a timer. I can see how people can just keep going here. How about we say everything explodes? Is it putting microwave food inside of the space suit? And now it's exploding and turning into cats in a space suit. Wait, what? More explosion. No. No, it's getting stuck. Oh, and there we are back with the explosions. Tacos everywhere. And we're cut off. Honestly, that was an experience. You could call this useless, but you could also call it fun. And you could also easily imagine a world where this is the worst it'll ever be. And maybe in the future it could like live life generate environments. I mean, heck, if you hook this up to voice recognition, you could just be speaking and it kind of visualizes everything you say. Maybe that plus VR plus gaming could make some really interesting experiences. But it is interesting and you could try it for free. And

obviously, it's a lot of fun. Let's see what's next. Okay, this story would usually be a quick hit, but I think it's actually a big deal and I want to tell you about it because UIO blocked out all of their users from downloading whatever they created on the app without notice. Apparently, they partnered with Universal Music Group. And I suppose there's a really clanky workaround to this if you have some songs in the account, but other than that, if you created music in there and you were hoping that it's stored within your account and you can just access it anytime, well, that just changed. you're not allowed to download your own songs that you created in there. And who knows, maybe other music creation apps or other AI apps may follow. So my recommendation would be if you have something critical or something that you created with any AI app, well, always download it and don't rely upon the accounts actually storing it forever cuz as you can see from this, well, probably first big example, which might change in the near future. I wouldn't rely on it, but it might. But the point here is that you don't actually own any of the stuff as long as it's in your account. So, make sure to download it because you might just get locked out of that without notice at any point in time.

Okay, let's move on. All right, so next up we have this week's quick hits. Not that many, but these are all the stories that I consider worth sharing. Starting with a brand new 3D model. We've seen many of these before, and they're always incremental improvements. So, now we have Seed 3D 1. 0 from Bite Dance. The demo does look quite good, but you do need a Chinese phone number to actually

use this. Another quick story would be Netflix going all in on AI for

advertising. And also similarly, I thought this story was interesting with all the meta people that OpenAI hired. Apparently, there's like a entire Slack channel of meta people at OpenAI and they're bringing a lot of the aggressive growth tactics over from Meta into OpenAI and they're exploring use cases like using memories for personalized ads is something that CEO Sam Alman previously called a dystopian idea. Anyway, this is just a new story for now, but things are moving quickly and as OpenAI is hiring a lot of people from Meta that are dissatisfied with the new researchers getting ridiculous comps, these people are moving companies and

getting creative. Then we also have Google Earth AI showing how they use AI in relation to Google Earth. So, it's a app that many businesses use to do weather forecasting for floods, wildfires, etc. And now they're adding AI tools so that these businesses can easily interact with it. So you can look for certain patterns all across the world just with natural language with

the power of AI. Now then we have yet another video model and I would call this a incremental improvement. Apparently it's good at dancing cuz all of the demo videos are relating to dance. It's high level 2. 3. You can generate four video clips for free every day. And that right there is pretty much everything we have for this week. I hope you found something that was interesting or inspirative to you. And with that being said, my name is Igor and I hope you have a wonderful week.