Ep 635: ChatGPT & Wal-Mart team up for AI shopping, Google drops Veo 3.1, Claude Skills get released & more

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.

ChatGPT teamed up with Walmart for AI shopping, but forgot to put the most important part in the cart. US President Donald Trump is facing new criticisms for how he's using AI. Video and Anthropic made some big moves inside of Claude, focusing on how people actually work. A lot happened in the world of AI news this past week and if you miss any of it, don't worry. Give me the next 30 minutes or so and I'll get you caught up on all the AI news that matters. What's going on, y'? All? Welcome to Everyday AI. My name is Jordan Wilson and Everyday AI is your daily live stream, podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with AI, but how we can make sense of it to get ahead to grow our companies and our careers. That's what you're trying to do. Fantastic. Starts here with the unedited, unscripted live stream podcast. But if you want to take it to the next level, if you haven't already, and why haven't you, go to our website, your everydayai.com go sign up for our free daily newsletter. There we recap each day's podcast, the highlights in the newsletter, as well as all of the other AI news that you need to know for that date. But this is our weekly segment. We do the AI news that matters on most Mondays. So if you're scratching your head all week being like, what the heck is happening? What should I be trying to do here in in my role with all these new AI tools and updates? Well, just tune in Mondays. We give you the no nonsense guide on what actually matters. So live stream audience, great to see you. So Bronson joining us from the YouTube machine, Brian and Jay on LinkedIn. Thanks for tuning in. We appreciate that. Let's get straight into our first news story. Yeah, OpenAI and Walmart forgot the most important thing in the cart. So Open OpenAI and Walmart partnered for a new instant checkout feature that allows Walmart users to search and buy products directly in ChatGPT. So Walmart announced its integration with ChatGPT this past week, allowing shoppers to purchase items from anything from towels to iPhone cases straight from Walmart's catalog within ChatGPT. But according to the Wall Street Journal, Walmart's Chat GPT partnership currently excludes fresh groceries, despite Walmart being America's Largest grocer. Yeah, so, you know, get the whole experience. And this limitation is pretty notable because ChatGPT excels at things like meal planning. So this could be a boon when and if this feature does become available. And Walmart's vast grocery business, including cold chain logistics and same day pickup and delivery, make it uniquely positioned to solve the challenge of free, fresh food, e commerce. But it's not available, at least now inside of ChatGPT. So industry experts say whoever controls the AI interface for shopping will control the future of retail, making Walmart's move inside of Chat GPT a pretty strategic one. So we'll see what ultimately happens with this new feature inside of ChatGPT, the instant checkout. And if I'm being honest, a lot of people haven't even seen it. So right now, if you're wondering like, how do I do this? You only do it right now with natural language. So if you're shopping for something. So I haven't even been able to trigger this one for Walmart or the, the other big one for Shopify or Etsy. So it seems like you really have to be shopping for something in particular. But I do believe once OpenAI rolls this out to more accounts, they get more retailers on board. I'm guessing there would probably be a dedicated mode that you can call on to make sure that you're always shopping when you want to be shopping. But I mean, just the, the food integration, right? I know so many people use Chat GBT to not just like meal plan, but to be like, hey, what can I make? So it seems like a no brainer for OpenAI to eventually work with the grocery side of Walmart. So pretty big omission, I think, you know, with this, the meeting of the two titans, right? The biggest AI provider in ChatGPT and the biggest grocer in Walmart. And you can't do the one thing that probably a lot of people thought you could. All right, but speaking of things that were actually very useful, this week Anthropic had a handful of pretty useful updates, and I'd say the biggest one is its new Skills feature. So Anthropic has launched a new feature called Skills inside of Claude AI and Claude Code, aiming to streamline how the chatbot performs specific work related tasks. So what the heck are Skills? Well, they're essentially folders that contain documents, instructions and potentially code allowing Claude to access tailored information for a given task. So right now the feature is only available to paid subscribers. Though luckily this time Anthropic did not just Limit it to its, you know, 100 or or 200amonth max plan. So normal Pro subscribers and team and enterprise users will have access to Claude skills. So here's how it works. There's some pre built skills right now inside of Claude that Anthropic kind of uploaded, but also users can create their own custom skills by uploading files and guidelines, enabling Claude to follow company specific workflows and branding requirements. So in these skills, Claude can automatically select and use relevant skills such as brand guidelines or design templates to complete tasks like creating pitch decks or posters. And then once created, skills are available across all Claude interfaces, including the web app, Claude code and the API, as long as they are enabled. So Anthropic has also built those standard skills for common tasks. But the ability to create custom skills is designed to help teams save time and maintain consistency. The skills feature also moves Claude a little closer to agentic behavior, allowing it to handle repetitive or complex tasks without needing repeated instructions. Yeah, live stream audience I'm curious if anyone has used skills yet. I've toyed around with them a little. Here's in my brief experience so far, here's how I would describe skills, because it's a. I wouldn't say it's confusing, but it's constantly changing, like capabilities are constantly changing. So one new capability inside Claude is the ability for it to create different file types, which is actually pretty big because it's something that large language models as a whole have struggled with. And it makes sense that companies like Claude and OpenAI should be focusing on this because ultimately people want to be creating different file types. So a couple months ago, Claude rolled out the ability to create files. So skills kind of also can take advantage of that. So I like to think of skills, at least right now, as something a little more advanced than a GPT. So if you're using a custom GPT inside ChatGPT as an example, you can upload custom instructions, you can upload files, maybe examples of outputs, but you can't really use all of these different modes, right? This canvas Mode inside chat GPT or you know, deep research. So GPTs are kind of limited. So when you look at skills, I look at them as a more useful kind of version of a GPT that can actually do way more things because you can not only, you know, put together, you know, instructions, but also you can take advantage of more of Claude's capabilities. So it's probably something, I don't know, maybe. We'll do it Wednesday on our AI at Work Wednesday series So make sure you tune into the newsletter. We'll probably put a poll out to see if you guys want that. Speaking of Claude, like I said, they had a couple big updates this weekend. Their second one here on today's show, maybe even bigger, especially if you're an enterprise organization. So Claude has launched a new integration with Microsoft 365 allowing organizations to search and analyze data across their enterprise documents, emails, calendars and chat tools all from one place. So Claude can now connect directly with Microsoft 365 including SharePoint, OneDrive, Outlook and Teams, making it possible to search and analyze documents, emails and conversations without manual uploads. So now kind of on par with where ChatGPT is with its Microsoft 365 connectors. So users can ask questions Claude questions and receive answers that pull from all the connected company data sources, not just their own files, offering a more complete view of the organizational knowledge. So the new enterprise search feature is especially useful for onboarding new employees, quickly finding company policies and identifying internal experts on specific topics. Well how does it work? How do you enable it? You got to be an administrator. That's always going to be one of the downfalls of getting these things rolled out across the organization is you always have this question is like who the heck is our admin? Is it this guy from 20 years ago who's been gone and no one knows where he is? So it's one of those things you're going to have to chase down your Microsoft 365 admin, but admins need to enable it on the Microsoft side and on the Claude side, the connector for their organization, after which individual users can then authenticate their accounts to for more seamless access. But once set up, every member of the organization automatically gains access to a dedicated company branded project that centralizes knowledge and custom search prompts. So this is, it's actually been just the last couple of months that we finally seen, you know, bringing in dynamic enterprise data into front end AI chatbots. So, so I, I will say it started with this connectors phase and now it's going a step further. So I haven't been able to test this just yet. Been a little busy, I'm actually on the road, had a trading that I was doing for a company the last few days so I haven't gotten a chance to pull fully put this through its paces. But between the the big shift towards connectors bringing in dynamic data, these Microsoft 365 direct integrations with both anthropic and obviously we've had it for a while with OpenAI, but as well as Google. You can't forget Google just released their version of Gemini Enterprise and Gemini Business, which kind of surprisingly shipped out of the box with Microsoft functionality. So it's kind of fascinating to see this and also be like, well, why did Microsoft even allow this? Maybe they're seeing, you know, users kind of shy away from using Co Pilot natively inside Inside Windows. Or maybe they're just saying, hey, if we want to remain as relevant as possible in the future, we need to let people access Microsoft data from whatever AI chatbot they use, even if it's not Microsoft Co Pilot. So bold, bold strategy from from Microsoft, so we'll see how it pays off in the coming years. Speaking of bold strategies and maybe a little weird, OpenAI is planning to allow adult content on ChatGPT amid a safety push so OpenAI CEO Sam Altman announced that ChatGPT will soon permit adult verified content, including erotica, for verified adult users, following some major upgrades to mental health safety tools, and this move could roll out as soon as December. So in a tweet, Altman said that OpenAI has completed work to better handle mental health issues, paving the way for relaxing restrictions on content for adult users. So all men stated that ChatGPT will be able to respond with more personality and fulfill previously blocked requests, including erotica for verified adults, and this obviously took the Internet by storm the past few days. OpenAI maintains that safety for children and teens remains a top priority with some of these age prediction technology that they're kind of rolling out to help bar those under 18 from maybe accessing more adult features. US lawmakers are responding to these changes, with US Senator Josh Howley circulating a draft bill to ban AI companions for minors. Entrepreneur Mark Cuban criticized this erotica move from OpenAI, warning that it could backfire and potentially harm young people, especially as the company faces scrutiny over its handling of youth mental health. And the adult content shift comes as other AI platforms, mainly Elon Musk's Grok xai, have already offered more permissive experiences, including sexual role play in their AI chatbot. Meta has faced controversy for chatbots that engage in sexual banter with minors. That was a gross one we have to cover a couple months ago, fueling concerns over AI safety for young users. OpenAI is also developing a separate ChatGPT experience specifically tailored for teens, following research and lawsuits highlighting risk to adolescent mental health. This one's a sticky one, right? It's sticky because a lot of users want this. A lot of more adult users. I don't personally understand this, but I understand kind of the free speech or freedom of expression take on this one. Right. Essentially. And this is kind of what, you know, OpenAI and Sam Altman said online. They said, hey, we need to let adults be adults and not necessarily police what they want to do inside of an AI chat. But yet at the same time, we know, I mean, look at just the history of the, the Internet and social media and how I think overwhelmingly toxic it's been for younger people. So I don't know ultimately how OpenAI is going to police this. And mainly, like, the thing I think of is shared computers. Right. If there's a family computer and maybe the, the main ChatGPT account is an adult account. Right. A verified adult account. You probably don't want minors using some of these features. Right? It's already out there. It's, it's the open Internet. I get it. So I don't know. We'll obviously be covering this in the coming months, but yeah, live stream audience, what's, what's your take on this? I get letting adults be adults, but safety of children is paramount. All right, before we get going, we have a lot more to cover. First, a quick word from our sponsors.

This podcast is supported by Google. Hey, folks, Steven Johnson here, co founder of NotebookLM. As an author, I've always been obsessed with how software could help organize ideas and make connections. So we built NotebookLM as an AI first tool for anyone trying to make sense of complex information. Upload your documents, and NotebookLM instantly becomes your personal expert, uncovering insights and helping you brainstorm. Try it@notebooklm.google.com.

All right, our next piece of AI news. US President Donald Trump again in some hot water from critics over how he's using AI generated videos. So the White House drew national attention by releasing multiple AI generated videos over the weekend in response to millions of people protesting Donald Trump's policies. And then President Trump shared some interesting and very controversial AI generated videos on social media following the widespread no Kings protest across the U.S. many called the AI videos distasteful and called into question how AI should be used by politicians against their political rivals. In one of the videos, Trump wears a crown while piloting a fighter jet labeled King. Trump dumping what appears to be feces on a crowd of protesters. Below, another video shows Trump in a crown and robe, wielding a sword as a song with the words Hail the King plays in the background, and political opponents, including Nancy Pelosi, are depicted bowing before him. The White House also posted an AI generated image of Trump and Vice President JD Vance wearing crowns alongside House Minority Leader Hakeem Jeffries and Senate Minority Leader Chuck Schumer. Democrats in sombreros captioned where built different critics called the videos inflammatory and accused the administration of mocking peaceful demonstrators exercising their constitutional rights. Well, you know, here we go. I think we went a whole one or two weeks without having to talk about one of these AI videos from US President Donald Trump, but it appears to be part of his ongoing effort to, I don't know, promote his campaign and silence critics, I guess. All right, next piece of AI news. This one, pretty big one it might seem a minor tweak on how you could use the Internet, but I personally think this one's big. So according to the information, OpenAI is pitching a new sign in with ChatGPT feature to companies right now, actively trying to get companies on board, aiming to make ChatGPT accounts the gateway for users to access AI powered features on third party websites. So this approach allows businesses ultimately to transfer the cost of using OpenAI's models directly onto consumers. So presumably this would make it easier for some companies to use AI features. So kind of lowering the level of entry to having certain AI features on their website, but also at the same time not having to handle the billing side of it or charging customers. Instead, customers can sign, or could in theory sign in with their ChatGPT account and then access certain AI features on these other websites. So we've been talking about this for a couple of months. It's been widely reported, but it looks like according to the information, we're getting closer to maybe this becoming a reality. So here's how it would kind of work. Well, users on the end would sign into partner sites using their Chat GPT account, similar to how you can log into certain websites using a Google or Facebook account. But the key difference, API usage costs would be deducted from the user's ChatGPT account and then free tier ChatGPT users would have their interaction with these third party sites count toward their existing usage limits, potentially causing free ChatGPT users who might use this feature to reach their cap more quickly and urge them to become a paid subscriber as well. This strategy is designed to help businesses lower their expenses while pushing more consumers toward a paid ChatGPT plan. And the move follows OpenAI's recent launch of apps in ChatGPT, which lets users interact with services like Zillow, Spotify, Figma, and Canva right now directly in the chatbot interface. And then the new sign on feature is part of OpenAI's broader effort to make ChatGPT a central hub for a growing number of online experiences. Here's what this ultimately is. I think this is being positioned and reported on as, hey, let's make it easier for businesses to bring some AI functionality to their websites, to their consumers. And at the same time, yeah, OpenAI might be able to convert people to a higher tier of plan. I don't think it's that, I don't think it's that. Let's, let's, let's cut this straight. What it is, it's about data, it's about personalization, right? The difference is this technology has been around and used on, I would guess, hundreds of thousands of websites, right? I probably use my Google account to log into hundreds of different websites so you don't have to sign up for an account. But generally when you do that, at least historically, before the age of generative AI, you're limiting these third party services to a very strict set of data, right? As an example, inside Google, I have everything, right? I have my emails, I have my Google Drive, I have Google Docs, I have all these, you know, financials and spreadsheets and videos and all these things. But when I'm signing up for these services, for the most part I'm just granting them access as an example, to read my email if that's, you know, a functionality of the end service that I'm using. I don't think it's going to be like this with the sign in with ChatGPT. I think when you sign in with ChatGPT, I think essentially you're going to be sending all of your conversational history obviously to ChatGPT, which could go in many different directions. So I don't think that this is a play ultimately making it easier for businesses to have AI service and you know, pass the cost off to the consumer. I think it's ultimately about getting more data, more training from companies because especially let's say you're on an E commerce site, right? That's some high intent searching and some high intent chatting that you may be doing with an AI inside there, logging in with ChatGPT, right? So imagine there was a Chat GPT esque experience, right, how you might chat with a chatbot, an intercom, a drift chat bot on so many of these websites. Imagine if it was just like a chat GPT bot, but specific to the website you were using. Well now all of a sudden that website has a ton more data on you. But also I do think a lot of it is probably going to pass on to OpenAI. So I will definitely, once this is start rolling out to third party websites, be taking a deeper look at what data is ultimately shared because I think it's going to be a lot more than people might realize. All right, our next piece of AI news. Google has just released VO 3.1, a significant update to its AI powered video platform. And it's already, I think, kind of evening the playing field between Google's latest AI video generator and OpenAI's Sora 2, which when it came out was I think heads and shoulders above VO3. So now with VO 3.1 I think it's a much closer fight. So let's talk a little bit about what's new inside VO 3.1 and V3.1 fast. So Google expanded those models with its AI video generation capabilities. So now in 3.1 it's available in Gemini, the Vertex AI platform and Google's Flow tool, which is really good. We haven't really done much on the Flow tool. Maybe we should start especially with some of these new features with VO 3.1. So the new model features richer audio and improved narrative comprehension. That's the thing I've probably been most blown away by. Right? There's yes, native audio syncing so you can make characters or yourself say anything. So the new model features a richer audio and improved narrative comprehension, meaning AI generated videos should sound better and make more sense as complete story. Also, there's enhanced realism in a true to life textures as a key upgrade and aiming to make AI videos look more natural and convincing, so looking more natural and convincing and sounding more natural and convincing as well. So the ingredients to video features now let users upload reference images and also generates videos with matching audio. There's also the new feature of Scene Extension which allows users to extend current video clips and that now supports audio in the Gemini API and Google Flow, but not right now in the Gemini app. Also, the last kind of new feature is first and last frame. Pretty cool. This allows users obviously to upload two different static images, one as a starting frame, one as an ending frame. And then Google VO 3.1 will fill in everything else in between, including in most cases with all audio as well as seamless transitions, especially if your first frame is very much different than your last frame. So live stream audience, podcast audience, let me know if we should cover these Google 3.1 or sorry, v 3.1 updates. I never know, right? It's. It's one of those things. I have a creative background I worked in, you know, marketing and advertising for, you know, a good 10 plus years. I, I don't know if if that's something that the average, you know, everyday AI listener is really, you know, wanting to learn more about. I do think maybe two years ago I would say probably not, but now maybe right. Especially for smaller businesses. I think before AI, video tools really just help creators be more creative. I think now where they are in the ease of use and simplicity, now they're helping non creative people be creative. So maybe it's something we'll focus more on again. I'll probably put out a poll on the podcast or in the newsletter asking you all about that. All right, Microsoft, Some news from Microsoft. They didn't let Claude and OpenAI and Gemini take all the fun. So Microsoft is rolling out a major update to Windows 11, introducing its co pilot AI as an always listening AI assistant for users who opt in. So this new Copilot features allows users to activate voice commands with the command hey Copilot. Similar to Apple's Siri, Amazon's Alexa and Google's Assistant. So yeah, maybe finally more smart assistants that aren't actually just dumb. However, this new feature is not enabled by default. Users must turn it on in the app's settings settings to allow voice activation. And once activated, a microphone icon and a chime signal will come on telling you that Copilot is listening for commands. And then to end a conversation, users can either say goodbye, click on the X interface on their screen, or wait for Copilot to disengage after inactivity. So Copilot can perform different Windows tasks, answer questions, and interact with Office apps like PowerPoint and Word using natural language. That's going to be the big one and how well it actually works and well when access rolls out. Right now Microsoft is also the new Copilot driven Windows Taskbar. Can't forget about that. And Copilot Actions will also let users book restaurants or order groceries from the desktop. So yeah, it's not just the new hey Copilot. There is also vision has been rolled out to more accounts as well as the new Copilot Actions. Some of these things that have been available for many months on the web version of Copilot starting to roll out across the desktop version. And the voice enabled Copilot will first be rolled out to users in the Windows Insider program starting in May and will expand to more users in the coming months. All right, our last big piece of AI news for the week. OpenAI. You know they've been Trying to say they're keeping up with compute a trillion dollars of partnerships. Well, now they're just, I'm gonna make their own AI chips. All right, so OpenAI is teaming up with Broadcom to design and deploy custom computer chips for their AI, aiming to roll out new AI accelerators late next year. So this partnership is expected to deliver computing power equivalent to 10 gigawatts, which OpenAI CEO Sam Altman describes as a gigantic amount of computing infrastructure for advanced AI. OpenAI's move to create its own AI chips signals a push to control the hardware side and future capabilities, potentially reducing reliance on outside suppliers like Microsoft. Broadcom shares surged more than 9% following the deal. Underlying investor excitement about the partnership. The collaboration though, began more than 18 months ago, according to Sam Altman. And Broadcom already works with some of the biggest name in tech like Amazon and Google on similar AI projects. OpenAI has recently struck up some huge deals with Nvidia and AMD for specialized chips, as well as data center partnerships with Oracle and coreweave. So yeah, not only are they working with AMD and Nvidia for specialized AI chips, well, now they're just getting in on the action themselves with this chip partnership with amd. And many of these deals though, involve circular financing, which has caused a lot of these AI bubble talks to keep bubbling. And this is where companies Both invest in OpenAI and supply it with technology, raising concerns about, yeah, this bubble is going to burst. This AI bubble is going to burst. All right, that's not all. There's a lot more that happened. Let's go over quickly some of the what's new and what's next. Yeah, a lot of other, I think, impactful AI news stories, updates and a couple of rumors that maybe didn't make our top stories this week, but let's go ahead and go over them. So Notebook LM released new video styles powered by Nano banana. Extremely impressive. WPP and Google partnered on a $400 million AI marketing innovation. Microsoft is prepping to add Sora to video generation to its online version of EcoPilot. Speaking of NotebookLM, more NotebookLM. They may be dropping slide generation updates soon. My gosh. Turning into the do everything platform. Love NotebookLM. Salesforce expanded AI partnerships with OpenAI, Anthropic and Google that they announced at their conference this past week. SORA extended capabilities to allow video generations inside Sora 2 for up to 15 seconds for all members and 25 seconds for ChatGPT Pro users. ChatGPT added a new chain of thought log on the right hand side of the interface. Anthropic, obviously we talked about this, but worth mentioning again, released a Microsoft 365 MCP connector. Microsoft started rolling out its own connectors online. Right now they're rolling out like Gmail connectors if you're using the online version of Copilot. I saw that and I'm like, this is a timeline I didn't expect, right? Google Gemini supporting Microsoft. Microsoft supporting Google. All right, and that is our last kind of what's new and what's next for the week. So we covered a ton, y'. All. Let me quickly recap our big news stories. First, Walmart joined Chat GPT and their new instant checkout program, but fresh groceries not included. Anthropic introduced Claude skills for enhanced AI workflows. Speaking of Claude, they now integrate with Microsoft 365 for Unified Enterprise Search. OpenAI is planning to allow adult content on ChatGPT maybe as soon as December. The White House shared some controversial AI videos amid the no Kings protests here in the U.S. openAI is set to unveil a sign in with ChatGPT feature. According to the Information, Google has launched its VO 3.1 update with some new features for AI video creation. Microsoft unveiled the Always Listening Copilot AI for Windows 11, kind of their answer to Siri. And OpenAI and Broadcom announced a major AI chip partnership. All right, it's a ton. As always. If you missed anything or if you want to stay ahead, that's where you do it on our website. If you haven't already, please go to your Everyday AI dot com. Sign up for our free daily newsletter. Also, if you're listening on the podcast, thank you for your support. Please, if you haven't already, subscribe to the show on Apple Podcasts or or on Spotify. If you're listening on Spotify, I'm going to be dropping more Spotify polls. All right, so make sure you check out the show note. I show notes. I want to make sure that we hear from our Spotify audience a little bit more and you tell us what you want. So thank you for tuning in. I hope to see you back tomorrow and every day for more Everyday AI. Thanks, y'. All.

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.