Ep 778: Codex Goes Remote Control, claude Goes Small, notebookLM gets super Powers and 7 more AI features you Can’t skip out on

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.

Meet Firefly AI Assistant, now live in Adobe Firefly, the all-in-one creative AI studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere, Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution.

Maybe I'm getting old, but I hate trying to do real work from my phone. My wife makes fun of me. Something about me acting like a boomer, I don't know, but I need at least like 50 inches of monitors to feel actually productive. And I had high hopes when the Claude dispatch computer use came out, but it was clunky. My connection to it broke multiple times, and the actual computer use inside of Anthropic's Claude Cowork was slower than me trying to write out the transcript to this podcast left-handed. I'm right-handed, but regardless, I feel I don't even write anymore. But that may have changed now because we just got OpenAI's updated ChatGPT app, which includes remote control access to codecs, which my gosh, sirens blaring. I think this is one of the most anticipated small feature updates of the year that will make a big difference. And that's not all of the new AI that we got this week that you might have missed. Anthropic made a big play for small businesses. Microsoft is actually killing off some dedicated Copilot features, but replacing it with something else. And NotebookLM got a ridiculously powerful upgrade that I haven't seen a single person mention. Yeah, because unless you're literally reading our newsletter every day and going out and doing more research on your own, you can't keep up with every meaningful AI update or new feature that gets added. And let's be honest, at least when it comes to the big four— that's OpenAI, Microsoft Copilot, Anthropic Claude, and Google Gemini— there's new features every single week that change your workflows. So This is what we do on Fridays. We're doing a new-ish kind of show called Fresh Friday Features where we tell you here's what's new and here's how to use it. So on today's show, stick with me. This one is going to be fast because I actually got like a fake root canal cleaning. My mouth hurts. All right. So if I sound weird, I'm not AI and I'll probably go a little faster today, but stick with me for 20-ish minutes and you're going to learn why Anthropic just dropped Claude inside of QuickBooks and PayPal. You're going to know the new way to control your coding agent from anywhere, and you'll understand why Microsoft just killed off Copilot mode and what replaced it. All right, let's dive into it. Welcome, my name's Jordan and this is Everyday AI. If you're new here, this thing's for you. It's a daily unedited, unscripted livestream podcast and daily newsletter helping everyday business leaders like you and me keep up with the latest and greatest in AI. I tell you what matters, what doesn't, how to use it to grow your company and career. So if that's you, oh yeah, you're like, oh, that's what I'm trying to do. Cool. Me too. All right. We do it all here live together, but make sure you go to youreverydayai.com. There you can go sign up for our free daily newsletter. We're going to be recapping the highlights from today's show as well as all of the other hot AI updates you've got to know. All right. So let's get into it. Let's look live. Here we go. Let's start with this one. Big move for small businesses from Anthropic. They introduced Claude for small business. So, which I like this. All right, so obviously I come from, you know, I had another marketing company where we mainly worked with small and, you know, medium-sized businesses. So I like a dedicated kind of offering here from Anthropic. So here's what it is. It's Anthropic's first ever SMB-focused package. Uh, so it's built on Claude Cowork, and it's essentially— you might miss it— it's a little toggle inside of Claude Cowork that essentially installs. You install it, and then you can then connect, uh, Claude to QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and even Microsoft 365. So it also ships with 15 ready-to-run agentic workflows. And 15 reusable skills covering payroll planning, month-end close, cash flow forecasting, invoice chasing, lead triage, contract review, and tax pre— uh, tax season prep. So essentially a lot of those, um, you know, routine and, uh, you know, I, I won't say mundane, but yeah, mundane things that a lot of small business owners do, right? Even those things I just named off there, I'm like, I spend a lot of my time doing that, right? Running 2 small businesses. So here's who has access. So it is available right now to paid users who can run Cowork. So you have to have access to Cowork that's on the desktop. So that's team and enterprise plans. Also don't train on customer data by default. All right. And here is what— how Anthropic kind of markets it. They say that We are launching Claude for Small Business, a package of connectors and ready-to-run workflows that put Claude inside the tools small businesses depend on to help small business owners take full advantage of AI and cross off items on the to-do list. Small businesses account for 44% of the US GDP and employ nearly half of the private sector workforce, but their adoption of AI has lagged behind larger enterprises. Tools and training are rarely tailored to the way small business— small businesses operate, and as a result, their use often stops at the chat window. As part of our public benefit mission, we are committed to helping business owners harness AI more fully and effectively for their most important work. All right, dude, this is great. Don't get me wrong. My, my hot take, I don't know if you all have been following, uh, kind of the developer side and what Anthropic's been doing over the last couple of weeks. I don't know. If anything, to me, this is like, uh, Anthropic's trying to go for a little PR here. right? They've made a lot of developers mad with a lot of changes they've made with how they're, uh, essentially how their service wraps around others, some overage charging, all these crazy things. People are not happy. This is the, I, I'd say the lowest sentiment I've seen, uh, again, uh, towards Anthropic maybe ever over the last week. So seems like they're kind of splashing this in like, right? Like number one, I love it. Uh, so, you know, hats off to Anthropic for doing this, but I don't know, to me it's almost like they know people are mad at them. They see some of the momentum slipping away. So, you know, they're trying to go in there and put on their nice PR hat. Alright. So here's why it's useful. Well, the biggest thing is it closes the gap where small business AI usage, you know, typically stops at the chat window. Right? So what that means is by having these kind of prepackaged, you know, these prepackaged skills, essentially, you know, and these kind of ready-to-run agentic workflows, it will probably kind of close the gap for those small businesses. That maybe don't have the skill set or just don't know how to piece it all together. So right now every task is user-initiated and does require approval before anything sends, posts, or pays, which fits owners who can't afford a runaway agent. All right. So here's who's going to find it valuable. Well, I think if you're a solopreneur, entrepreneur, small business owner, you know, if you have like an under-50 employee company and you don't have a lot of dedicated specialists, right, especially if you have people out in the field. If you're someone that's doing it all or you have a, a, you know, small admin team, small marketing team that's stretched out thin, I think this is going to be extremely useful. One thing to keep in mind. Yeah, some of these prepackaged, you know, just speaking very truthfully here, some of these prepackaged flows, Claude is not always the most efficient in how it uses tools. All right, so I'll just say this. If you're on a base plan, a base paid plan, or even the team plan, and you're using these things and you're finding utility in them and you're using them a lot, I'm just letting you know, small business owners, you're going to go through a ton of usage. You're going to hit your, you're going to hit your rate limits very quickly. I'll just say that. So obviously this is a moneymaking move for Anthropic, but I do think overall it's good for small businesses. All right, next one. This one, I don't know if I saw literally anyone talk about this. So this dropped on Tuesday and I think it takes advantage, uh, of Google Workspace Studio, which might be Google's most underutilized kind of tool. So if you've ever used Zapier, Google Workspace Studio is essentially their version of Zapier if your business runs on Google Workspace, right? So if you have you know, most, most companies are either Microsoft companies or they're Google companies. So if you're a Google company, you know, ask your IT or, you know, your system admin if you guys have Google Workspace Studio. If not, get it right away. It's really cool. You can have all these, uh, kind of prebuilt flows. But now with this new feature, it's amazing. So here's what it is. There's a new Ask NotebookLM step inside of Google Workspace Studio, uh, and it lets you use existing NotebookLM notebooks as a grounded knowledge source for Studio automation flows, which is absolutely bonkers the ways that this can be used. And then it generates responses based on those insights, summaries, and research already in a notebook rather than an open web search or relying on Gemini's training data. So who has access to this right now? So it did already start a rollout, and in true Google fashion, we don't really know. It's just this slow rollout, but it has already started, right? So these Friday features, we only cover new AI features that are available. Just might not be available to a ton. So it's rolling out to rapid release and scheduled release domains starting this week with full visibility. Google did say by the end of the week, which is today. So if you didn't get it earlier this week, you might already have it. You do have to be on Business Starter, Standard, Plus, Enterprise, Plus, Education Fundamentals, right? All these, you know, Google AI Pro for Education. Google has so many tiers. I get it because they have like hundreds of millions of users. But yeah, just make sure if you are the admin that you default this on for Gemini is enabled for your entire workspace. Make sure you have alpha features enabled as well and make sure that you have Workspace Studio enabled for your organization. So here's what Google says about it. They say we're integrating Notebook LM into Google Workspace Studio. This enhancement lets you use your existing notebooks as an AI knowledge source for your automations. You can use the new Ask Notebook LM step to generate grounded responses on the insights, summaries, and research in your notebook. So, like I said, if you haven't used Google Workspace Studio, it's amazing. So in this example here that I'm showing on my screen, they have a very simple 3-step flow that they're showing. So step 1 is when you get an email and then you can set conditions. And in this instance, they're doing a return policy. So you can set a condition to where if you get an email and you can use Google to determine with natural language if someone is asking about a return policy, and then if they are, as an example, you can insert a NotebookLM notebook in there as the source to answer those questions. That's extremely important because NotebookLM still has a ridiculously unfair advantage over literally everyone, OpenAI, Anthropic, Microsoft, everyone, because they're the only ones that I think has this, like, grounding piece figure it out because even if you use a project in ChatGPT, in Claude, even in Google Gemini, right, it can still pull in training data and not just the data that you upload. So the grounding in NotebookLM is amazing. It's only going to pull answers from what you get it. I've seen like, I don't know, I don't have a scientific, you know, number, but I'd say like a less than 1% hallucination rate, which is so, so good. So if you haven't used this, you should check it out. So here's why it's useful. Well, it closes the gap between NotebookLM as a research surface and then Workspace Studio as an automation surface. So your flows can cite your own sources. So this replaces kind of an Open Gemini step, right? So this is what I used before, what a lot of people that were using Workspace Studio used, right? And then the downside with that is, well, it can— it's gonna pull from its own training data, which could be bad, it could be good, but then it can also pull from the wider web. So this is just a way to constrain those responses and make sure that it stays in the lane of the data that you already have improved. There's so many different use cases for this. So here's who's gonna find it valuable. Well, I think really anyone, but I mean, if you're in enablement, uh, in enablement ops, knowledge management, if you're an educator, learning and development, marketing, uh, customer service, customer experience, whatever, uh, it's huge. So, uh, this already has started to roll out. Um, but I do think it's going to be fully rolled out in June. So yeah, make sure if you haven't already, check out that you have access and have all those features enabled. All right. Next, real-time voice from OpenAI. This is a big one. And here's why. The voice model from OpenAI, the default one in ChatGPT, hasn't really been updated since GPT-4.0. This new version is not in there yet, but OpenAI did say that they're bringing it to ChatGPT, which is why I think it's extremely important to talk about right now because right now it's just on the API side, but it is rolling out. So here's what's new. This is OpenAI's new flagship voice model with what the company calls GPT-5 Class Reasoning. So it is a voice, a real-time voice model that reasons. So this was also launched alongside GPT Real-Time Translate that translates between 70+ input languages and 13 output languages, and also GPT Real-Time Whisper, which is streaming speech-to-text. Yes, streaming. This also has a 128K context window, which is 4 times the context window of the old Realtime Voice, which is GPT Realtime 1.5. And then you also have adjustable reasoning efforts, uh, between low, high, and extra high. So if you do want to build something like, like this on the backend, uh, it's really easy. I've actually been, uh, you know, toying around with this in Codex. You can just build yourself a personal, you know, real-time voice agent, you know, a super— right? I love Codex. So now I'm kind of building with this. Kind of a real-time voice model that can access all of my data in real time. So who has access right now? Well, it's live in the OpenAI real-time API. So as long as you are using the backend of OpenAI's dev platform, you have access to it. It's also reachable right now. I know this is a little more on the dev side, but it's reachable via the WebRTC, WebSocket, SIP. It also supports the OpenAI Agents SDK and also remote MCP servers, which is big one, uh, right now. Um, I don't know if it's fully rolled out in the EU yet. Uh, so here's why it's useful. Well, it collapses the old transcribe plus reasoning plus synthesizing stack that used to happen, uh, into a single audio-in audio-out model with reasoning. That's the big thing, right? Because the previous voice model, even the advanced voice model, did not have reasoning, and it wasn't honestly very Good. Now, this one is really good. Actually, their demo, I highly recommend you watch it. We did share it in the newsletter last week, so I'll make sure to share it again. So yeah, make sure you go sign up for the free daily newsletter at youreverydayai.com and make sure to, under the GPT Realtime 2, make sure to watch this demo because it is extremely impressive. So here's who's going to find it valuable. Well, any voice agent developers shipping phone-based customer support, sales qualification, in-app voice flows. If you're working around anything with voice, real-time voice agents, this is huge. Also companies running global support that previously stitched together transcription plus translation plus text-to-speech, right? So if you were using multiple, you know, kind of models before, which was the standard way to do it, especially earlier in, you know, early 2025 and, you know, before that, it was standard to have to stitch together 2 to 3 different models, right? One to transcribe, one to think and reason, one to convert it back to text, right? So this just brings it all into one model. So here is what OpenAI says about it. They said, we've introduced 3 audio models in the API that unlock a new class of voice apps for developers. With these models, developers can build voice experiences that feel more natural, respond more intelligently, and take action in real time. So here they say GPT Realtime 2 is our first voice model with GPT-5 class reasoning that can handle harder requests and carry the conversations forward naturally. GPT Realtime Translate, a new live translation model that translates speech from 70+ input languages to 13 output languages while keeping pace with the speaker. And GPT Realtime Whisper, a new streaming speech-to-text, uh, that transcribes speech live as the speaker Talks. All right. And there's actually a real cool demo, uh, on that page as well that you can actually just do it in real time. So yeah, this thing, if you don't know how to use it, literally just go into Codex, you know, point it to this page and say, hey, I want to, I want to use this. So you have to, you know, set up your API key and everything like that. But once you do that, you can literally just talk to Codex and have it build you a personal voice assistant, right? That's something if I, you know, didn't spend 10 hours a day working on this podcast, I would already have an amazing one up and running. All right. Uh, We have a lot more. Everything from what the heck did Microsoft do with Copilot mode to new updates with Claude, Code, and Codex. But before we get to all those, let's take a very quick break for a word from our partners.

Adobe just introduced an entirely new way to create, bringing the power and precision of its Creative Suite into one conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app. The all-in-one creative AI studio powered by Adobe's Creative Agent. Firefly AI Assistant lets you start with your vision, just describe what you want and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows drawing on 60+ pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with Creative Skills, a growing library of with pre-built workflows for common creative tasks like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible, so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI Assistant, now in public beta. See it today at firefly.adobe.com.

All right. Our next Friday feature that I don't want you to miss, uh, again, a little more technical, but if you've ever wanted to build something and you weren't really sure how, this one is going to be for you. Yes, you need to be using Codex for this one. Don't worry, go listen to Wednesday's episode. I covered a beginner's guide for Codex. That's episode 776. So there were like 2 and a half updates that impact Codex. So I go recommend you listen to that, but Our next one is the OpenAI Developers plugin for Codex. So this is a Codex plugin that connects Codex to the OpenAI platform. So it's 3 bundled capabilities. So it connects Codex to platform.openai.com to create and wire up a project API key from inside Codex. So like what I was just talking about, you know, using the real-time voice model, now it's even easier to do something like that inside Codex. Also, you can diagnose common OpenAI API errors. And this also works together with the bundled OpenAI Doc skills that ships with Codex. So who has access? Well, anyone using the Codex app, right? So you just need to go look in the plugins and then look for OpenAI developer plugin, and then you're off to the races. And here's why it's useful and who's going to find it valuable. Well, it just removes you having to go back and forth between the Codex environment if you are trying to build something and just connect directly to OpenAI's platform. This makes building something easier than ever because, yeah, a lot of times this is what I would be doing, right? I would have the, the platform, right? I think it's like platform.openai.com open and, you know, managing your API keys and looking up documentation. Now it's literally all just built into this plugin. Right? So when OpenAI says Codex is the super app, they mean it, right? I'm even starting to use, uh, the built-in browser more and more. Uh, I'm using like ChatGPT, right? This isn't that meta, uh, but inside the Codex app, it's actually a great browser. That's where I use ChatGPT because I have Codex use ChatGPT for me because if I know I need to run a series of, you know, multiple prompts or, you know, multiple GPTs or multiple workspace agents, right? If I have to do things manually, I just have Codex do it and I say, all right, go do this and wait for 2 minutes. Anyways, here's why it's useful. So it can invoke it automatically when a task involves an OpenAI API setup. So once you have it linked up first, you know, building an agent, you know, from a one-line prompt is so easy because it can actually wire up all your credentials. And then it also adds first-party troubleshooting for API errors. That's huge. So you don't have to go you know, looking, you know, I know people don't really go to Stack Overflow much anymore, but you don't have to ask, you know, open a new tab and ask ChatGPT why something isn't working and, you know, screenshotting things because it's all literally just built in. So I think, you know, the people who are going to find this most valuable, developers who are prototyping apps and agents, you know, non-technical people who just want to build something and they're, you know, not sure where to start. This is great. Also teams using Codex as their default agentic harness who want OpenAI platform actions to live alongside their plugins, they're going to find it useful as well. All right, next, what the heck happened to Microsoft Edge? So they kind of killed off Copilot mode, but most of the features are still going to live on. But yeah, Copilot mode is, at least as a dedicated mode, is officially gone inside of the Microsoft Edge browser. So Microsoft is retiring bringing the standalone Copilot mode in its Edge browser and folding those AI features directly into the default browser on desktop and mobile. So here's a couple of things that are added and things that are new, things that aren't there anymore. So mobile gets near parity with desktop. So that's nice. Now there's not like two very different versions. And maybe that's one of the reasons why they killed the official Copilot mode is to bring more parody if you are using Edge on desktop and Edge on mobile. So you also on mobile now get the multi-tab reasoning that you previously had in Copilot mode in the Edge desktop browser. So essentially you get all these great mobile updates, right? You get the new, the journeys, the long-term memory across past chats, which is really good. And you also get vision and voice, which is where Copilot can see and understand what's on your your screen. So I mean, just right there, right? A lot of people are— there's some things I think a lot of people don't like, like the history in the browser is not as clean as it once was because it's a little more conversational now. So there's some things that people don't like about this update. But I think from a mobile perspective, if you're not married to a mobile browser right now, which I'm not, right? I saw this update, I have Edge on my phone, and I'm like, okay, well, maybe I need to be using it a little bit more. So Google has slowly rolled out some more AI in you know, agentic features in Chrome browser, but it's kind of been a slow rollout. So, you know, at least when I saw this news, it's like, okay, this is high on my list of starting to do this week. And then there's also some new desktop-only additions, including the new Study and Learn mode, which unfortunately ChatGPT got rid of. Uh, and then that turns open tabs into guided study sessions and quizzes. There's also a new writing assistant and a turn open tabs into a podcast feature, which I'm going to be checking that one out as well. So who has access? So it's rolling out now across Edge on desktop, iOS, and Android over the coming weeks. So yeah, these phased rollouts, right? When I say these things, they are available to some and there's no, you know, rhyme or reason. Sometimes you'll get them right away. I've seen some people have gotten the new version of Edge. Some people haven't. So just make sure you open both the desktop and the mobile version, check for updates, you know, log in, log out. I know on the enterprise side, obviously a lot of companies use Edge, so this might be one of those worth checking out. So it does remove that mental task of switching between a regular browser and a separate AI mode. So it's just kind of available everywhere. So you did have to enable Copilot mode previously, so now it's just always on. And also bringing that multi-tab reasoning to mobile as well is great. In the long-term memory, let's Copilot pick up the research, shopping, or trip planning or whatever you're trying to do. So who's going to find it useful? I mean, anyone, right? I'd say anyone if you're not married to a current browser, especially a mobile browser. I'm a very heavy Chrome user. It is worth noting that Edge is built on Chromium. So your Chrome extensions and all of your Chrome data can sync over. So that's important to know. And yeah, so I think at least the big benefit here is getting more mobile AI features sooner. Because those unfortunately are a little bit slower to roll out to Google Chrome. All right. We have two more. We'll go through them here quick. Next is Claude Code taking a page out of the Codex playbook, and they brought goal mode to the command line version of Claude Code. So this is not the desktop version, right? The Claude desktop. This is if you're using Claude Code in the command line. Interface. So it is a new command in Claude Code that was released this week that sets a completion condition and keeps Claude working across turns until the condition is met. Right. So instead of giving Claude a prompt, well, you just give it a condition to meet. So this has obviously been a very popular feature in Codex, the command line version, and the OpenAI team did confirm that they are bringing this to the desktop version. So right now it's not available in Claude Desktop, or in Codex Desktop, but it is now available officially in both of the command line tools. So if you are more on the technical side, you know, this is pretty good. So essentially, after every turn, there's a small fast evaluator model. So Claude Code uses Haiku by default that reads the conversation and decides yes or no on whether the condition actually holds, right? And then goal auto clears when met. It can be cleared with with a /goal clear and works in interactive mode, headless mode, the desktop app. It says the desktop app, but I didn't see it working on the desktop app and also remote control. So who has access to this? So yeah, if you're on the latest version of Claude Code, you have to be on a paid plan and you should have access. So here's why it's useful. Well, it separates the worker from the judge, right? So the model writing the code isn't the one declaring get done, right? And that session keeps going to a fresh evaluator each turn, and it eliminates the, you know, kind of the common check-ins that you have, right? The keep going, or did you finish, or, you know, make this better, right? It, it's kind of like a, a simplified version of the Ralph Loop that was very popular, you know, 3 to 6 months ago. Um, and probably we're going to talk more about goals in the future, uh, as they start rolling out to all the desktop apps as well, uh, because I actually think that goal planning in the short term, right, until we get these fully autonomous agents, which we're not there yet, right? We have agents that can schedule, but when you get the goal command, I think that is the next big step toward fully autonomous agents, right? And you do have to change, I think, how you use and how you prompt these systems with the goal in mind versus giving them a command. And it does require, you know, more oversight and just refining your skills a little bit more because let's be honest, it's not very hard hard or even time-consuming always to get an agentic model or an agent off and running or scheduled. What it does take a little bit of practice doing is getting it to properly work toward a goal, right? That's where your language and, you know, kind of those, you know, quote unquote old school prompting skills from 2023 really start to pay off, right? When working toward a goal, because, you know, it's one of those things, if you're, you know, the saying is if you're one degree off your goal, eventually that ship is going to be lost at sea. And I think the same thing can be said for these new /go commands. So like I said, Codex was first to this in the command line interface. Now Cloud Code following suit. But regardless, the reception to this one hasn't been as positive. Most people— I haven't used this one yet in Cloud Code, but most people have said that this one's a little clunky. It feels rushed. That's not me. That's just the overall sentiment so far. all right. And then last but not least, this is the big one. Codex in ChatGPT. Now they are starting to get married. All right, so this was just announced yesterday afternoon and Codex is now live inside of the ChatGPT mobile app. So your phone can now connect to and control a Codex instance running on another machine. So this is, OpenAI says it's more than just a remote control for a single task as the mobile app app loads the live state of a connected machine and lets you work across all active threads, approvals, plugins, and project contexts. Um, and it updates, uh, updates the stream to the phone in real time, including screenshots, terminal outputs, diffs if you're looking at differences between, you know, one, one output and another, test results and approval prompts, files, credentials, permissions, and local setup stay on the host machine. So So this is big. And I do, I do want to scroll through here in this video. This is a little newer here, but I do believe I'm just checking here. I think you can control. I should have, I should have checked this before, before starting, but I actually think yes. Okay. So the thing here, which I absolutely love, is you can control multiple computers with this. So for people, if you are an entrepreneur, solopreneur, or just someone with multiple computers like me, I've talked about this a lot. I, for the most part, use 3 separate Macs, um, on a day-to-day basis. So this one for me is big because, uh, unfortunately the Dispatch, um, the Dispatch version, uh, on Claude, so it's the Claude iOS app, uh, controlling the Claude desktop app. The Dispatch goes to Claude Cowork, but it doesn't go to Claude Chat. It doesn't go to Claude Code. So it's a little disjointed and mine was always breaking. It's still broken. I had to literally uninstall, uh, Claude and, you know, all this. But this is one thing I love in Codex aside from it's just one, uh, you know, one interface. There's not a separate chat, cowork, code, right? It's just one. But you can control multiple computers if you are using multiple computers with the same account. So that's great. I can literally be off, I don't know, in the Costco parking lot and I can be running things on my laptop and running things on my Mac Studio. So pretty cool. So who has access? Well, if you have, well, everyone has access, right? Because Codex, even though the limits aren't great on a free plan, right? If you have a basic paid plan, you're going to get the same limits as you would on $100 a month. You know, the $20 a month ChatGPT plan is as good or better in terms of limits than the $100 a month Anthropic Max plan. So if you have a free or a paid plan, just make sure to check in your, uh, ChatGPT app. So this isn't a Codex app, uh, that's coming out to iOS. This is just the normal ChatGPT app. What's weirdest to me is it didn't— unless it updated in the background, I didn't even have to update it. It was just there. And then you can pair, uh, then you can pair it, and then you're off to the race. Uh, you're off to the races. So, uh, Codex on Windows though is not yet supported, but OpenAI did say support for that is coming soon. So here's why it's useful and who's going to find it valuable. Well, it lets you approve and run Codex nonstop. And I think a lot of people like I talked about on Wednesday's show, make sure to go listen to that 776. I think it was one of my better shows I've maybe ever done because I think it's that important, right? OpenAI has said Codex is the new super app. So for nearly 900 million people who are using ChatGPT, yeah, you gotta start getting familiar in Codex. But it lets you approve anything in Codex and control a machine completely. It keeps those long-running agentic workflows moving across the day without forcing you to babysit the desktop. Right. You saw all those, you know, funny memes online, you know, how people were leaving their, you know, laptops cracked open and just all the time, right, going from one place to another. I did it myself, too. But also it uses a secure layer so trusted machines stay reachable across devices without being exposed to the public internet and active session states sync anywhere you're signed in with ChatGPT. So I think engineers are obviously going to love this if you're doing long-running tasks. Teams who are already developing inside of managed remote environments via remote SSH, which is a new feature, well, now you just get phone-side approval flows on top of those environments. Also, if you just have, right, the great thing about Codex is, well, it works like a human. Its computer use is unmatched. It can control, you know, your Chrome on desktop. you know, there's so many things you can read and write to any folder that you give it access to. So literally at any time it can open up other apps on your computer, it can open up other websites, it can do anything that you can do, right? If you sit down and work with it. So it is a general knowledge working behemoth all in one, right? Where unfortunately Anthropic, it's 3 different silos that don't have memory. Codex is very different. It is much more powerful. So now you get all that on the go. For me, this is the thing I've been most looking forward to, not gonna lie, uh, in like 3 months, right? Because there's been a lot of rumors of this, so it's finally here. I think it's by far, uh, the biggest update. Here is what OpenAI says about it. They said Codex is now in the ChatGPT mobile app, so you can stay in the loop from anywhere while Codex gets work done across your laptops, dev boxes, or remote environments. As agents take on longer-running work, a new rhythm for collaboration is emerging to keep work moving, you need to be able to easily answer a question, review what Codex found, change direction, approve what comes next, or add a new idea. More than 4 million people now use Codex every week, and we're seeing how much those small moments matter. A quick check-in can keep a thread moving, prevent unnecessary rework, or help Codex make progress with the right context. Now you can do that all from your phone. All right, this is a great one. All right. So that's a wrap for this week's Fresh Friday features. If you missed any of these, don't worry. I just filled you in. We're also going to be recapping them all in today's newsletter. So make sure you go to youreverydayai.com and sign up for that. So if this was helpful, do me a favor, takes 30 seconds. If you could leave us a review on Spotify or Apple Podcasts, I'd appreciate that. Make sure you also follow or subscribe to the show that helps other non-technical business leaders like you cut through all the AI fluff, cut through the marketing, and just get the answers that they need to grow their company and their career. So thank you for tuning in. I hope to see you back next week and every day for more Everyday AI. Thanks, y'all.

Meet Firefly AI Assistant, now live in Adobe Firefly, the all-in-one creative AI studio. Just describe what you want to create in your own words, and the assistant handles the rest. Orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stay in control with the ability to step in and refine at any time. See it today at firefly.adobe.com.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us.

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.