Ep 768: Microsoft Outlook Goes Agentic, Gemini Gets File Creation, Manus brings Cloud computer and More. 7 New Practical AI Upgrades

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. While Anthropic and OpenAI had relatively quiet weeks in terms of new features, Google and Microsoft rolled out some very small updates inside of Gemini and Copilot that will actually have a very big difference in your workflow. If you're someone who's constantly unimpressed by AI slides, you might want to check out what Replit just released. And Manus may be taking a shot at Perplexity and OpenClaw with their latest update that gives you an always-on virtual computer that you can pretty much claw around with. So yeah, there's a lot more that was released this week in AI. And if you miss anything, if you don't have time to spend hours a day testing out new features or just seeing what's being released, don't worry, I do that for you. And on Friday, we have our new segment called Friday Features to keep you up to date. So on today's show, here's just a sampling of what we're going to go into. You're going to learn how you can finally tame your inbox the right way with Microsoft Copilot. You're going to know the smallest new Gemini feature that makes a big difference, and you'll why and how a vibe coding platform might make a dent in slide creation. All right, let's get to it. Welcome to Everyday AI. My name's Jordan Wilson, and if you're new here, we do this every day. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with everything that's happening in AI, but how we can use all this information to get ahead, to grow our company and our career. So if that's what you're trying to do, it starts here with the unedited, unscripted livestream podcast, but to become the smartest person in AI, make sure to go to our website. That's your cheat code, youreverydayai.com. All right. And in today's newsletter, we're going to have all the other AI news because, yeah, FYI, we are just because some travel, I'm having to record this on Thursday afternoon. And I already know there's going to be at least one new update that I'm going to miss. The Codex team said they are or hinted that they're dropping something. Probably not going to get to it on today's show unless they drop in the next 20-some minutes. But Thursday is a big day for AI updates. So we might actually miss one or two things that we would normally catch. So make sure to check out the newsletter. All right, let's get into it and let's just go ahead and bring up my screen here. So yeah, um, we don't really do anything overly visual on our Friday shows, so if you're on the podcast, don't worry, you're not really going to be missing anything too much. So first, let's start with Gemini. You can create docs now. In Gemini. Well, what's the big deal? Why does that matter? Well, it matters a ton because Anthropy was actually first to market with this feature. OpenAI followed somewhat later with their ability to update or to create documents in ChatGPT. And Gemini was kind of on an island for the last like 6 months and they were the only big player that you could not actually create documents inside of Gemini. But that has changed. That's because Google just announced that you can now inside Gemini generate complete downloadable files from a chat prompt without leaving the app. So here's some of the supported formats. It includes Google Docs, Sheets, Slides, PDF, Word. So, you know,.doc, Excel, CSV, LaTeX, Markdown, RTF, and plain text files. So this replaces the old manual workflow, which I was personally tired of doing, which is copying the text out of Gemini, pasting it in a different app, and then having to spend a ton of time reformatting, right? All of that annoying, you know, human duct tape that I talk about all the time. This is a big quality of life update. So who has access? So this is rolling out now to all Gemini users at no additional cost. So there are similar features, right? So if you've seen the Help Me Create in Google Docs as an example, this is separate from that. So I'm just talking about this is just inside of Gemini. So here's why it's useful. Well, you can probably already tell it's going to save you, especially if you are a heavy Gemini user and you're someone that's constantly having to produce some type of output and share it with your team, right? So if you're creating slides, sheets, documents, etc., this is huge. So it eliminates that nonstop and so annoying copy-paste reformat loop that was a main friction even for me, right, after having a great session inside of Gemini. So users can go from a conversation directly to a finished shareable file in a single step. So who's this gonna be valuable for? Well, literally anyone, right? So knowledge workers, students, teams, basically anyone that's already using Gemini. And like I said, you have to produce something on the backend. So this one is huge. Files, they can also be downloaded to your device or they can be exported directly to your Google Drive, which I like that because if I'm being honest, sometimes even in ChatGPT or in Claude, you know, yeah, I'll download it, but ultimately I just want it to live in my drive somewhere. So, you know, I think Anthropic has that, that capability. I don't believe ChatGPT does, but this is a nice feature from Gemini that you can either just download it or save it directly to your Drive. So big, like I said, small little footnote feature, right? And there's a good chance you miss this one. But if you are a heavy Gemini user, this is big, right? Even for me personally, I know that there are so many times, right, when I knew I have to create a document, I have to create, you know, something. And I know the end goal, it's, it's, oh, I have to attach this, I have to send this to someone, or I need a file to go into my Drive. And there's been multiple times in my head where I'm like, oh, I would like to do this in Gemini, but it's actually going to just create a bunch of extra work because I know I can't save that as a file, right? So at least for me, and I'm sure power Gemini users are going to really appreciate this because this was a pretty big shortcoming for Google, right? Especially the fact that, you know, the they weren't even the first with the Drive integration, right? So great quality of life update with this release. All right, next. And this one, y'all, speaking of like small things that make a big difference, finally we have Agent Mode inside of Microsoft Outlook. So yes, you can have a much smarter AI Copilot working in your inbox. So Microsoft launched Copilot Agent Mode. Through its Frontier Early Access program. So Copilot now runs continuously in the background. That's the key part here, the big update to manage your inbox and calendar rather than just responding to one-off prompts. So some new specific capabilities can include just kind of email triage and prioritization, drafting follow-ups for unreplied threads. Yeah, that's me. Get those lost all the time. Creating new inbox rules, resolving meeting conflicts, rebooking rooms, and blocking focus time. So yes, you do have to write now be in the Frontier Early Access Program to get access to this. But I know there are millions of users who do have access to that. So it's not like in a beta version or sign up, right? Many larger enterprise organizations are part of Microsoft's Frontier program. So this is rolling out to a lot of people. So aside, you obviously have to have Microsoft 365 Copilot license to use that as well. So right now the rollout covers Outlook for Windows, web, iOS, and Android for inbox features. And deeper calendar features are currently Outlook for Windows and web only. Yes. So EU users, sorry, not supported at launch, as with most things. So here's why it's useful. Well, it moves Copilot from working individually and you having to push the button to actively maintaining your inbox and calendar as an ongoing, you know, background job. And the context handoff is key, right? We talked about this last week kind of in our Start Here series, the concept of agentic context Carried. Here we go again. This is a great example of putting that into practice, right? So this is essentially Copilot can escalate, and then the full conversation context carries forward so you don't have to re-explain to Copilot. So this reduces the coordination drag by automating kind of those repetitive maintenance layers of office work, right? So this is big, and this is essentially a version of Google's AI mode inside of Gmail that hasn't rolled out broadly just yet. So kind of, you know, Google Soft and Microsoft or Google Soft, Google and Microsoft. Yeah, that would be a company. Who's making that one? Elon? Google and Microsoft kind of racing to bring this always-on agent that's thinking in the background and, you know, helping you triage your email, which is great. But I, I don't know, is it, is it too little and too late with these desktop agents, right? Especially when you have, you know, something like, you know, live artifacts in Claude, which I've been loving, which kind of can do this for you. Or even if you, if you just, you know, have, you know, something saved in canvas mode and ChatGPT, etc. But maybe for the non-technical people, the people that, you know, are just heavily ingrained in the 365 ecosystem, this is going to be big. So yeah, those enterprise knowledge workers, especially with high email volume, are going to find this one really, really useful, right? If you're constantly, you know, especially if you're already a heavy Copilot user inside of Outlook, you've probably experienced the frustrations of having to re-explain context or even just having to handle things one by one, right? Yeah, you've probably benefited from using Copilot inside of your inbox, but, you know, having it more as an agentic partner is pretty big. So some other things here. Microsoft did warn this is a preview and capabilities may change. Bulk actions may be incomplete and there's some pattern detection features that aren't yet available. All right. Here is what kind of how Microsoft marketed it. They said Outlook used to be where you worked. Starting today, it's where Copilot works for you. Until now, Copilot in Outlook helped with the task in front of you— drafting an email, catching up on a long thread, or finding a time to meet. Useful, but not the hardest part. The real work is everything around it— the follow-ups, the SIP, the slip, the messages that need attention, and the scheduled changes that pile up before the day even starts. That's what's changing today. Copilot in Outlook is now agentic, taking on the ongoing work of running your inbox and calendar. It triages emails, reschedules conflicts, and surfaces what matters most before you even ask. So pretty big update. Like I said, if you are a power Outlook user, which I know most enterprise, you know, people are. You have to be. All right, uh, next one, not really big news unless you are a heavy, uh, AWS or Amazon shop, but if you are that, you're going to love this. So, uh, this is now being released, the Amazon Quick desktop app. Uh, so Amazon launched, uh, their new native desktop app for Amazon Quick, which was formerly called Quick Suite. Uh, so this was just announced at the What's Next with AWS event. Uh, I think that was yesterday, 2 days ago. Uh, so the app runs— seeing a trend here? It— the app runs continuously in the background and it accesses local files directly without you having to upload them. Then it sends proactive operating system level notifications and can automate browser-based tasks and desktop applications. It also builds a personal knowledge graph that learns a user's people, projects, and relationships across every connected tool and session. So I mean, pretty cool. It's just, I don't know if I've met many people that are using Amazon Quick, but like I said, if you are a heavy AWS or Amazon organization, this one does sound like something, a new feature you should be checking out. So it is available right now in preview on both Mac and Windows in the US East. So yeah, it's rolling out regionally, which not, normally don't see that, but that's what AWS is going with here. Here's the other thing. No AWS account is required right now. So users, so individually, right, you can sign up with a personal email, Google, Apple, GitHub, or Amazon credentials. And there are free and plus pricing plans that were announced alongside with the desktop launch. So here's why it's useful. Unlike browser-based AI tools, it can access local files without uploading them and operates proactively without the users needing to prompt it. So the persistent knowledge graph compounds over time, meaning that context carriers— context carries across sessions rather than starting fresh every chat. So it does integrate with Google Workspace, Microsoft 365, you know, Salesforce, Slack, Teams, Zoom, some of the big platforms. So it almost to me seems like Amazon's version of something between a Copilot, right? Like a Microsoft Copilot that runs, you know, at the operating system level and it connects to all your major carriers, but also something kind of like an OpenClaw Lite. So I'm not exactly sure what they're kind of aiming for with this, but here's how they spin it. So they said Amazon Quick on your desktop. Your work lives in— your work lives across dozens of tools, Slack, Microsoft Outlook, Salesforce, spreadsheets on your desktop, and emails in Gmail. Qwik is an AI assistant for work that connects to all of them, learns what matters to you, and takes action on your behalf. Set it up in minutes. By the end of the day, you'll wonder how you ever worked without it. Right? So, uh, sounds right. I haven't been able to personally, uh, use this yet. Like I said, it is fresh out. I've been traveling, right? I was in San Francisco. I'm back in Chicago. I'm going to St. Louis. So I haven't had time to you know, test this one out yet, but actually I might, right? I haven't used very many of Amazon's products just because I'm very, you know, heavily focused on, you know, the big four in OpenAI, Anthropic, Google, and Microsoft. But, you know, we'll see. Maybe this thing gets some momentum behind it. Maybe they get a lot of updates. Maybe it becomes, you know, as everyone's looking for that kind of autonomous worker that works in the background that brings, you know, AI capabilities to your desktop, maybe this is something that will catch on. All right, next AI feature that you need to know for this week. Well, slides. What are you using for AI slides, by the way? Personally, I'm absolutely loving the new ChatGPT Images 2 for slide creation. We did a show on that on Wednesday, actually. I'm So good, right? There's so many choices though. You know, there's Gamma, you know, Canva's updated some of their capabilities. You know, obviously inside of NotebookLM using Nano Banana to create slides, freaking fantastic. But there's a new player and that is Replit. So Replit Slides. Replit just launched Slides, their AI-powered tool for creating presentation decks. So Users can start from scratch. They can use Replit to build templates or import existing PowerPoint files, and then the AI can restyle it via a PowerPoint to code converter. Then you have export options that include PowerPoint, PDF, and Google Slides. So who has access right now? So Open, uh, Replit's own changelog, uh, changelog indicated that slide deck functionality is live and it's being actively updated. But right now there's not a lot of information on who actually has access, but it looks like anyone with a Replit account. So right now, exact plan requirements aren't really confirmed. Like I said, I think this one also just came out like today. It's only been out for like 2 hours as the time of this recording. So I don't think everything's been updated on their website yet. Maybe I'll find out here as I refresh it and take a look at it. But here's why it's useful. Well, if you are already a, you know, power Replit user, right? So if you're doing anything on the Vibe Coding in, if you're, you know, using it for any sort of web app, mobile app, etc., it might be worth it, you know, if you're already in Replit's ecosystem. So it does also kind of bring in a new potential category of customers for Replit. So it extends Replit beyond just the dev, you know, dev tooling into more of, uh, productivity for non-coders. So I think this is great for, um, you know, PMs, marketers, and just your general business users. Uh, the feature that I kind of liked is kind of this PowerPoint to code converter, you know, means that users can just import an existing branded deck and then have Replit AI kind of restyle it rather than starting from scratch, which I like because I make ugly slides. Um, I still do a lot of them manually. In Canva for certain things, right? Especially if I have to stick to certain brand guidelines. So this might be somewhere I take kind of an ugly deck that I do, or maybe my daily slides, and I'll throw it in there to see if it can beautify it a little bit. But who's going to find it valuable? I think people like product managers, right? So kind of the crossover between people who would already be using Replit for like prototyping. And then, you know, if you, you know, demos over memos, right? So it might be something like that. I think product managers, you know, people who are marketers, entrepreneurs, solopreneurs that may already be using Replit. Uh, it's kind of like one of those things. Since you're there, since you have all your assets already living inside, uh, I think it's pretty good. Also, you know, non-technical business users who are looking maybe for an AI-native alternative to Canva or Gamma. So if you have kind of tried a couple of other, you know, AI-native slide tools and you haven't really found one, uh, you know, this one could be it. So, uh, yeah, let me see this. Let me just go ahead and refresh this, see if we have any, any new information. Uh, okay. So It does say start creating for free. So not a lot of information out yet, but here's kind of how Replit spins it. So they say turn content into slides instantly. You can drop in a doc or rough ideas and get a complete slide deck in minutes. They say get polished designs with zero effort. You can start with beautiful prebuilt templates or import existing branded templates. And then it says to tailor every detail to your needs. It says you can customize the number of slides, the slide order, slide templates, and more. Also, it says you can get structured storytelling by default. You can automatically organize your content into a clear narrative with the right flow. So, uh, yeah, looks like you can start using it for free. Um, yeah, so maybe at least for a short time, uh, it's free, and maybe we'll see some pricing. If so, we'll make sure to update, uh, you all on that in our newsletter. All right, next. Yeah, Deepseek V4. Oh my gosh, it's going to change the world. It didn't. You know, this actually came out like a couple hours after our last week Friday features. But given that there weren't a ton of technical, like technically model releases this week, decided to go ahead and cover it even though it's, you know, technically a little older. But I think it is important. So Deepseek launched Deepseek V4 as a preview with two different variations, V4 Pro and V4 Flash. So both are a mixture of expert models with a 1 million token context window and up to 384K max output. AI moves too fast to follow, but you're expected to keep up. Otherwise, your career or company might lag behind while AI-native competitors Leap ahead. But you don't have 10 hours a day to understand it all. That's what I do for you. But after 700+ episodes of Everyday AI, the most common questions I get is, where do I start? That's why we created the Start Here series, an ongoing podcast series of more than a dozen episodes you can listen to in order. It covers the AI basics for beginners and sharpens the skills of AI champions pushing their companies forward. In the ongoing series, we explain complex trends in simple language that you can turn into action. There's 3 ways to jump in. Number 1, go scroll back to the first one in episode 691. Number 2, tap the link in your show notes at any time for the Start Here series, or you can just go to starthereseries.com, which also gives you free access to our inner circle community where you can connect with other business leaders doing the same. The Start Here series will slow down the pace of AI so you can get ahead. So V4, V4 Pro is a 1.6 trillion parameter model where V4 Flash is a 284 billion parameter model. So both are open source under the MIT license with weights available on Hugging Face, and both do support thinking and non-thinking modes. So availability. It's already available, right? So it's available now via the API, via the, you know, the chat.deepseek.com. So you can use it via the chat, you can use it via the API, you can use it through OpenRouter. Like I said, you can download it if you got a beefy machine, you can download it, you know, via Hugging Face and run it locally. So the OpenWeights are freely downloadable for commercial use. So v4-flash is the more practical right, self-hosting target, you might have to have about 160-ish gigabytes of RAM. So yeah, you're going to have to have a beefy machine to run that one. So here's why it's useful. Well, it's priced well. So v4 Pro is priced at only $1.74 per million input tokens and $3.48 for million outputs. So who's going to find this valuable? Well, development teams who are running high-volume workloads where cost per token really matters more than absolute benchmark rank, uh, or organizations needing open-weight models for compliance, self-hosting, or on-prem deployment. So kind of have their, um, their benchmarks up here. So nothing really shocking, right, in terms of capabilities. Everyone thought this, uh, you know, Deepseek V4 was gonna change the world. You know, here's the thing, when the last version of Deepseek came out, uh, you know, and it kind of rattled the stock market, it was because Deepseek said, oh, it only costs, you know, I think they said like $5 million to train or something like that. It obviously didn't, uh, since, you know, all the big labs and the US government have accused Chinese companies like Deepseek of model distillation. I said all along, don't pay too much attention to Deepseek, right? It's very likely distilled according to, you know, the labs and the White House, uh, you know, so even this version benchmarks, it's, you know, it's a, you know, top 5 open source model, right? But it's not really going to be competing with anything on the proprietary side. And like I said, it's not even the best, you know, open source Chinese model right now. So probably not worth testing out, right? Aside from if you just like to test out the new models or if you were previously using a version of Deepseek for something else and it's a simple upgrade. Otherwise, I'd say probably pass on this one. I know Deepseek gets gets a lot of attention, right, from, uh, you know, what happened, uh, last year. But I don't know, at least for me, I'd say this one is probably a pass aside from just some basic testing. All right, next, Manus computer. Yeah, we have another entry into the Perplexity computer slash OpenClaw ecosystem. So, uh, this is another one hot off the presses. So just announced today, uh, Manus cloud computer is a persistent always-on a virtual machine that runs in the cloud and users can spin up to run bots, Python scripts, or software 24/7. So unlike the standard Manus sandbox, which resets after each session, the cloud computer retains files, installed tools, and context across sessions, functioning like a personal remote machine. So it's accessible via SSH or the Manus web dashboard terminal. So no coding required to use it Manus handles the configuration through natural language. So right now, who can get it? Well, it's available to all paid users on the web and mobile. That's the thing that I'm looking at. I'm like, okay, native mobile support. The tech side is already set up. I'm not a big OpenClaw person, just FYI, just because I don't have the time, right? I don't have even 5 extra hours a week to configure OpenClaw. Tested it out, got it running, and I'm like, not really for me. I'd rather just, you know, time is money when I think about it. So that's why something like, you know, Perplexity's computer, you know, Perplexity's personal computer, I'm more apt to be looking at those. So this one, at least for me personally, is pretty interesting, right? Anyone that handles, you know, kind of that sandboxing and giving you a virtual machine. I know there's a lot of ways that you can do that with OpenClaw as well. So to each their own. But if you're already a Manus subscriber like I am, this one might be worth checking it out. So right now there's 3 plans. There's a basic, which gives you simple Python scripts. There's a standard, which gives you active websites and APIs, and then advanced, which also gives you team databases. This is available now globally. So here's why it's useful. Well, it solves the core problem at least that every regular Manus chat, right, starting from a blank slate. So the cloud computer lets Manus build on the previous work, run live databases, and maintain persistent knowledge bases across weeks and months. Also, you can have bots running 24/7, right? Slack, Discord, WhatsApp, right? So whether you're using them for customer service or just directing them to go do work without users need to configure AWS or manage server keys or write any infrastructure code. So who's going to find this valuable? Well, I think non-technical users, right? If you are kind of tempted by the, you know, having that virtual computer that you can run and it can be running for you or around the clock, you know, small business owners that want the bots, the database and the scheduled workflows running without a dev on staff or just people who are currently power users of Minis but have maybe been limited by that session-based sandbox and want more persistent compounding work. So one thing to note, the cloud computer can't access your local machine files, right? So it's fully isolated, which some people might see as a downside. For me, that's kind of what I want, right? So I think this is the alternative, you know, versus buying a dedicated computer, you know, for OpenClaw or buying a dedicated computer for Perplexity Personal Computer. So this is more of a, I'd say, a direct competitor, a much earlier version for perplexity computer, right, which runs in their cloud instance. So here's how Manus says spins it. So they say, introducing Cloud Computer from Manus, a dedicated machine in the cloud that runs your bots, Python scripts, and software around the clock. Now anyone can build and run anything. So they say, why use a cloud computer? A cloud computer gives you two things your laptop can't. It never turns off. Your files and setup stay put. So we will have this in today's newsletter if you missed it. So make sure to check that out. All right, our last big feature for the week, we have live real-time voice agents inside of Microsoft Copilot Studio. All right, let me bring up the correct one here. There we go. So here's what it is. Microsoft announced general availability of real-time voice agents in Copilot Studio, launching initially through Dynamics 365 365 Contact Center. So this is big, right? If you are in customer service, if you use Dynamics 365 Contact Center, which I know a lot of bigger organizations do, especially if you're selling products, this one's big. So there are— these are speech-to-speech agents that are optimized for low latency and natural interruption, meaning callers can speak normally, change direction mid-call, and then the agent adapts in real time rather than following rigid you know, scripts. So yeah, if your company handles a lot of inbound calls, whether it's for sales, customer service, something else, maybe you've been on the fence about should we do an AI voice agent? Maybe you've tried a third-party platform, but you're a Microsoft shop. In this case, this is going to be for you. Also, a call can escalate to a human agent, and then the full conversation context carries forward automatically. So the human customer support person doesn't have to repeat themselves. So So who has access? Well, it's generally available now in Dynamics 365 Contact Center customers in North America. Customers outside of North America must allow cross-geo processing since the real-time voice AI model is hosted in North America only as of now. So yeah, EU, not getting it, sorry. All right, so here's why it's useful. So traditional IVR and scripted bots, well, they break Now, if a caller interrupts or if, you know, someone gets frustrated or if you combine multiple requests into one sentence, right? Traditional, you know, AI chatbots can get screwed up a little bit. So the integration with existing Copilot Studio infrastructure is big, right? That means IT can apply the same governance, permissions, and compliance controls to voice as they do to text agents. So yeah, I think this is a small maybe subset of people out there, but if you are already set up in Microsoft Copilot Studio and if your company, you know, handles a big amount of voice calls, this is going to be big. Microsoft says that over 80% of Fortune 500 companies already have active agents in Copilot Studio, giving this a large ready-made deployment base. So who's going to find this valuable? Well, any contact center, uh, leader in North America. If you're looking to reduce call handling costs and improve first call resolution, or any customer experience or customer service teams already on Dynamics 365 who want voice AI without replacing their entire existing stack. All right, so that's a wrap, y'all. Uh, a lot of smaller updates this week, week, which means I think we're going to be in for a big week next week. So yeah, OpenAI and Anthropic were kind of quiet. Like I said, I do expect a Codex release probably in the next 24 hours, so make sure to check out today's newsletter since this is technically a prerecorded one. We did see some actually meaningful security rollouts for both both ChatGPT and Claude, as well as enterprise customers. But, you know, at least for today's show, I wanted to focus on actual new AI features that you all can go use today. So I hope this one is helpful. Like I said, maybe a little quieter on the feature end. All that means is we had a little calm before the storm. We obviously have Google's I/O conference coming up in a couple of weeks. We know Anthropic and OpenAI are going to be cooking around that. Microsoft, like I said, bringing more agentic flows. So it's an exciting time if you follow AI to be catching up on new features on Friday. So thanks for joining us. If you haven't already, please go to youreverydayai.com, sign up for the free daily newsletter. Thanks for tuning in. We'll see you back next time 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 youreverydayai.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.