Ep 626: ChatGPT’s new Agent Builder, Apps and 5 BIG takeaways you probably missed

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.

Will this be the update that finally makes agents mainstream? I think so. If you missed it, OpenAI announced a straight up cornucopia of AI updates at their Dev Day conference. And I know you might be thinking, I'm not a developer, so why does this pertain to me? Well, I think what OpenAI announced is going to actually reshape how we all work heading into 2026. I think it's going to be so impactful, it's actually going to shake up some legacy industries. So if you don't want to be left behind, stick around for the next 25ish minutes as I break it all down for you. All right, I'm excited for today's show. I hope you are too. If you're new here. Welcome to Everyday AI. My name is Jordan Wilson and I'm the host. And this thing, it's for you. It's your daily live stream, podcast and free daily newsletter helping everyday business leaders like you and me not just keep up, but get ahead. Because AI updates are happening all the time and I understand it's hard to keep up with them. So this is how you do it. It starts here with the live stream podcast. But if you want to take it to the next level, be the smartest person in AI at your company, you got to go to our website at your everyday AI dot com. There we're going to recap the highlights of today's show as well as keeping you up to date with everything else happening in the world of AI news. Let's make this one a fast one. Let's just get straight into it. ChatGPT has a new agent builder, Drag and Drop. You don't have to be a developer. And that's why I think even though this is called Dev Day, OpenAI's yearly conference that they hold in San Francisco primarily, it's usually been for developers. And even though it's still called Dev Day, I don't think it was actually for developers anymore. So on today's show I'm going to go over some of the show stopping announcements that I thought two of them, agent kit and apps. I'm going to talk about the improvements OpenAI made to the API and how that will impact ecosystem, even non developers. Just your average everyday AI enthusiast. And I'm going to unveil my five biggest takeaways from the day that I think Most people missed. So there was a ton. We're going to be sharing all of the updates in our newsletter today and we did drop quite a few of them yesterday as well. But a lot of the updates kept rolling out throughout the day. Let's start with the first one and maybe the one that most non developers may end up using the most because it actually was kind of a two pronged announcement. OpenAI announced apps in chat GPT. So this is kind of, well, developers can obviously make them, but we all use them. Think of your iPhone, think of how many apps that you use. Or your smartphone, think of how many apps that you use. Right? So this is OpenAI, both open, opening it up for developers, but also for everyone that uses ChatGPT and OpenAI actually with another small announcement. They said now they have 800 million weekly active users. Where it was just in September, I believe, where they said they had 700 million active weekly active users. So let's talk about apps in ChatGPT and then I'll talk briefly about what the heck is the difference between apps and GPTs. I'm still trying to figure it out, but it looks like GPTs might just be left to die in the long run. So apps right now, well, you can use them to work within your conversations and they work inside of Chat GPT. So chatgpt can suggest an app right now when you're talking to it, when it thinks that one could be helpful and it will simply call an app. You can simply call an app by its name to start using it, or you can select it from a dropdown. So right now the way that it works is you have to connect these apps one by one, at least the ones I've been using. The few of them that are available now, there's only a handful of them. But the end goal is for, well, when there's a lot more apps to ChatGPT, just to kind of seamlessly blend in app usage within your normal conversation. So as an example, using interactive features like maps or presentations. So the first apps that are available now. So yeah, you can go in, log into your chatgpt if you have a paid account and you can connect them. Now I've already connected Canva, Spotify and Figma myself. There's also Zillow that's available now. OpenAI does say that your privacy is protected with a permission prompt before the first use. I did see that. And then after that it doesn't come up anymore. And essentially apps, the big thing, the big difference is it uses your conversations context so you can be chatting about different types of things. The example that OpenAI gave is they were, you know, creating a pitch deck. So they were creating in Canva and then they were saying, hey, what new locations might our startup operate? And then it use Zillow to, you know, find housing in that location. That was just, you know, one example, but a pretty good one about how you can carry context over from app to app, which is actually pretty nice. I mean, again, imagine if you could do that on your smartphone, right? If you didn't have to, you know, copy and paste things between different apps or if your, you know, if your email app just knew what was happening in your text messaging app and it gave you suggestions, right, that would be great. So that's kind of the concept here in ChatGPT, using the power that is generative AI and large language models, not just to be able to help you create more useful content, but to be able to use the context and then to be able to work inside of those apps. And they did say that more partners, as an example, like Pellet, Baton and Uber are going to be coming later this year. They gave the example of Coursera and the big difference with apps and GPTs. If you're confused, I'm still trying to figure it out a little bit. I'm sure more clarity will come. But in my dealings so far, it's more of an interactive experience using these apps, right, where you can use these GPTs and you can go create a GPT within chat GPT. It's usually more text based, right, when you're interacting with a GPT. Yeah, even Canva has a GPT. Right? But it doesn't create anything visually in the app. Essentially you're working with text and you can create GPTs inside of ChatGPT. At least right now you can't create apps inside of ChatGPT. It is more of a developer tool for developers to create these apps. And OpenAI did say that they will be accessible, accepting developer submissions soon. So I don't know if this is going to be, you know, quite the same as plugins originally were or GPTs. And I'm also interested in what's going to happen with GPTs. I personally hope they stick around, but we'll see. So what does this mean? Well, like I said, developers can start building their own apps today. And OpenAI did say that they'll have a dedicated directory for browsing and searching for apps. And then like I said, once you add the apps, OpenAI will suggest them and surface them throughout the conversation. OpenAI will soon share details on how they can monetize their apps, so we'll see if that ever comes to fruition. Right. We heard the same thing when OpenAI announced and really hyped the GPT Store. And I didn't really hear or see anyone getting paid off of the GPTs. So I don't know, maybe someone did reach out to me. If I'm wrong, I didn't really see anyone. But I do think this is different because like I said, it is a richer experience. So you know, another example is they gave Coursera, right? I made a course on a top ready course on Coursera on generative AI like three years ago. So you could watch that inside of Chat GBT but go through the actual video modules and still have some space on the screen. Right? So what they're doing here is they're really just bringing the web experience in my, you know, dealing with these apps so far into Chat gbt. So just making it more of like an operating system. So apps will also become available for business and education users later. Now the big one grabbing headlines. Agent Kit. So the Agent Builder is actually part of Agent Kit, but I think the Agent Builder is actually what got so much of the attention from Dev day and I think rightfully so because it's a visual drag and drop builder, right? I'm actually wasn't planning on doing this, but I'm actually going to log into my OpenAI account right now, my Builder account, and I'll show you this right now. So if you're listening on the podcast, don't worry, you can always watch the video version on our website at your everyday AI dot com. So just sharing my, my screen here. This is the Agent Builder, right? You don't really have to know anything. It's a drag and drop interface. You can add, add and delete nodes. You can, you know, select different models here. So if you've ever used anything like Zapier, N8N any, you know, visual builder, that's all it is, right? And it's really powerful that you can build an agent backed and powered by OpenAI's most powerful models, right? You can use GPT5, you can even use GPT5 Pro. You can use GPT5 codecs, right? So you can build a lot of different agents for just about any use case using the Agent Kit. All right, let's quickly talk about it. So it's Agent Kit is essentially a toolkit for helping even non technical people build helpful AI agents or AI assistants. It simplifies the process of creating, launching and managing agents and businesses can go launch it now. So you do have to have essentially an API account set up or a developer account set up. So you're not going to find this on chatgpt.com you actually go to platform.OpenAI.com you also do keep an eye on this. You do have to set up billing, right. I did this, did this obviously years ago. So if you're going to be using these in production, right, they're not free, you know, to test them out, it's not going to cost you anything. But if you're ultimately using these in production, you know, you're going to be paying for the API usage. And that's the big difference. And the big differentiator is, yeah, you can go put these on your website or you can go put these in a web app. And that's the big one of the big differences aside from just being able to create agents and agentic workflows on demand without a lot of tech skills. So like I said, the Agent Builder itself is a visual builder that allows you to design AI workflows using simple drag and drop technology. Makes it faster and much easier, right, than you would have to if you were, you know, staring at a screen trying to MacGyver some Python strings together. And then you can easily add that once you're done and once you can test it, so you can test it in the actual platform there. So like I said, platform OpenAI.com and then go to the Agent Builder so you can test it out there, but then you can literally just have it live on your website. And OpenAI has some pretty clear and easy instructions on how you can do that. The cool thing is, is then you can also manage all of your connectors inside chat GPT and all of your data sources from within one one central admin panel. So it's not like you have to go, you know, manually build, you know, data pipelines. That's where the ease and simplicity of ChatGPT really shines here with the Agent Builder inside of the Agent Kit, there's also new evaluation tools to help test and measure the AI agent's performance. And it also doing it that way allows developers to fix and identify issues way quicker. There's also an open source safety layer that helps protect AI agents made inside of Agent Kit from misbehaving, according to OpenAI. And also there's some specialized training options, guardrails, some other things that you know, when you think of agentic AI or AI agents, I think we always think about capabilities, we don't think about hallucinations, we don't think about, you know, guard rails. So it's nice that in the first, at least I wouldn't say this is the first iteration, right, because we've had the OpenAI agents SDK now for, I don't know, about half a year. So this is nice that the first, more consumer, you know, first kind of non developer version of this has that built in. There's also some specialized training options that are available to make the AI smarter. There's also some templates that you can choose from when you go into the Agent Builder. So very easy, clean. If you've ever, like I said, if you've ever used anything like Zapier N8N to build agentic workflows or whatever, you're gonna find this very simple to use and not a huge learning curve at all. All right, before we get going, and hey, we might actually make 25 minutes this time, let's quickly pause for a word from our partners. 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, the next update at Dev day was kind of technically a smaller one, but some bullet points on what's new. Codex, which is essentially the coding kind of IDE from OpenAI, got technically official. It became generally available even though it was previously available. And this helps programmers write and review computer code way quicker. A bunch of companies, you know, OpenAI gave some use cases, but they said Cisco as an example, used it to cut their review times in half. Side note, we're actually going to have the president of Cisco on the show next week, so make sure to tune in for that. Speaking of Cisco, also, OpenAI said that the daily usage of Codex has grown more than 10 times since August. I believe Codex was originally introduced back in in May. I believe at the time it was only for pro users. So now it's just a much wider and kind of GA push here. But some new things inside Codex that OpenAI did announce, now you can ask Codex for help directly inside of Slack, developers can add Codex's powerful abilities into their own applications as well. And like I said, even in the Agent Builder, one of the Models that you can use is GPT5 Codex, which is a specialized version of OpenAI's most powerful model, built specifically for coding. There's also a new dashboard that helps managers track usage and monitor the AI's work. And administrators can now enforce safer defaults for their teams. All right, the next announcement. Kind of a big one. Sora 2 is now available in the API. So if you're like, what the heck is that? Well, you have probably heard about Sora 2, the ultra viral, but also in my opinion, kind of brain rot, you know, tool that OpenAI pushed out. Don't get me wrong, it's fantastic. But in general, I don't know, I'm not a big fan of social media. I'm not a fan of, you know, endless doom scrolling. So that's kind of unfortunately how OpenAI rolled out Sora. It is an iOS app. You have to sign up for it via the iOS app. It was invite only and essentially you just. It's only AI videos and it's just constant scrolling anyways. But the actual model itself is absolutely bonkers. Good. It is wild to believe that it's real. Anyways, now that's rolling out to anyone. I would guess that probably by week's end there's going to be dozens of third party AI video platforms that are going to be using Sora's. Sora 2's API. So as an example, I'm not really as big into the AI video scene as others, but I saw companies like Higgs Field VEED, right. So a lot of these AI video companies are now, well, you can use Sora 2. Well, why would you want to do that? Well, there's a lot of other features in these other platforms that maybe right now if you're like, oh, I wish Sora 2 had blank. Well, you might be able to get that now in these other AI video tools now that Sora 2 is available via the API. And I did already see within hours more than, I think more than a dozen different AI video platforms that had support for Sora 2. On the API side, OpenAI said that it is 1080p, so HD resolution and 20 second clips. If you don't know, I mean sort to absolutely crazy. The temporal consistency, character consistency from shot to shot, you know, audio syncing, soundscape, sound effects, it's understanding of physics. Really, really good. We did cover Sora the other day. If you want to go listen or watch that episode, maybe Watch is better. It is episode 623. All right. A couple other API updates just Going to go over these quickly. All right, so GPT Image 1 mini is now available in the API as is GPT5 mini pro. So previously developers could use GPT5 but not the most powerful version of it, which is GPT5Pro. That one is available in the API and got a pretty good hoot and holler from those in attendance at Dev Day as well as a new and what OpenAI said was an improved voice model in GPT Real Time Mini. So what this means again, same thing, probably between these three updates, hundreds or maybe thousands of apps that you maybe use all day, maybe that your, your company, your business might use. Probably good chance that it might be powered now or in the future by one of these three things that I just talked about. So let's. Okay, I got, I got my, my bullet points on the other screen here. So, so with GPT5 Pro, like I said, now it's API accessible. The most advanced language model OpenAI has for higher accuracy long context reasoning in complex tasks with GPT Real Time Mini. This is a new updated version of advanced Voice mode with lower latency API voice model enabling real time streaming audio for live assistance, also at a way lower cost. So, well, if you need something cheaper or faster, you know, maybe your company wants to provide a voice assistant on your website for your customers. Well, you could have used the old GPT Real Time, but now the GPT Real Time Mini is a little faster and way cheaper. And OpenAI says from a quality standpoint, not a big drop off. And then last but not least, you know, all those, you know, super viral AI images that people create all the time in chat gbt. Well now they're making a new version of that available for developers as well in GPT image 1 mini. Compact image generation API model for fast cost efficient creation and editing. All right, that's what actually happened on the Surface, right? You don't have to go watch the 90 minute keynote now. I just told you what you needed to know for the most part. But I do want to talk about now five takeaways that you probably missed. All right, so reading between the lines a little bit here, some things that were said at Dev Day and after. This one was actually after, during a Q A session after Dev Day where Sam Altman said that profitability is not in the top 10 concerns for OpenAI. I also said this like a year ago. I said OpenAI is not worried about profitability in the same way Anthropic is. And I said a while ago that's one of the Reasons why I think OpenAI will ultimately beat Anthropic. And he did say that right now their strategy is a space of investment in growth with confidence they'll build a profitable, profitable model eventually. I love seeing Sam Altman say, like he said on his personal blog the other day, that they have to figure out how to make money with Sora, right, Because it's very expensive. But I think what they're doing here, reading between the lines, they're playing the extreme long game, right? They're not necessarily worried about making money off the average, you know, Chat GPT plus user. And another thing I've been saying for a long time, the fact that you, we can get most of this technology. Yes, I know some of the shinier bells and, or louder bells and shinier whistles, you know, might not be available on that 20amonth plan. You might go to the 200amonth plan, but still, just, just the capabilities of these models. It is a. I mean, we're all stealing, right? Essentially. But what is happening here? They're. They care about dominance, they care about users, right? That's one of the things that they're pushing all the time. You know, you always see these reports about Anthropic's revenue going up. You don't see as many of those about OpenAI. One of the reasons is they're bleeding cash. But yeah, I don't think Anthropic is ever going to be able to say 800 million weekly active users like Chat GPT just said, right? OpenAI's playing the long game here. And I think taking a page out of Amazon's playbook, right? There's so many things in Amazon's, you know, famed story from bookseller to you can't live without them, that they just lost money along so much of the route just because they were winning users, they were winning trust, they're winning confidence and they're winning the ecosystem, they're winning the platform. That's what OpenAI's doing here. So one big takeaway there. Next one. This is interesting. This event is, you know, mainly for developers. It's. And these developers are. Well, they're running startups for the most part. What OpenAI just released kind of cannibalized a lot of them. And this happens a lot of times with OpenAI. And it's always kind of strange for me, right, when you see these large AI companies, you know, they work with developers to get feedback to make their product better and then ultimately this release comes out and they wipe out everything, right? Drawing On Amazon, right, they, they've been accused of kind of doing the same thing. They see which products are selling and then Amazon makes their own version, exact version of that product and undercuts them. So same strategy here. I don't know if there's eventually going to be any uproar in the developer community, but like literally Agent Kit in the Agent Builder, I've seen hundreds, and I'm not exaggerating when I say hundreds, hundreds of AI startups that essentially they're just using, or they were using Open AI's agents SDK and just providing essentially a UI for OpenAI's technology. And then they might say, oh, it's for marketers, right, but you're just throwing in a system prompt into UI and just using OpenAI's technology anyways. So OpenAI kind of just squashed a ton of startups and it's going to continue to happen. So I'm wondering if at any point there is going to be in uprising. The other thing that's interesting is I think that with the Agent Builder, they're taking a direct shot at zapier@make.com even kind of workado, you know, it's a huge, this AI automation industry, a huge $8 billion market. All right, number three, GPT5, Codex, I think kind of represents the beginning of the end of traditional programming. And you can say this whether you believe it or not. OpenAI said themselves that nearly every OpenAI PR goes through a Codex review and that OpenAI engineers complete 70% more pull requests per week when they're building with Codex. Right. So it's pretty telling that open, even OpenAI said that the majority of their new code that's created is created by Codex. And, and you know, they are setting the industry standard. Same thing with Anthropic. Anthropic said something similar. If you're not seeing this, you know, if you watch the entire Dev day announcement or if you read about it and if you're not being like, yeah, this is the beginning of the end of traditional programming. I'm not saying it's the end, it's the beginning of the end and why, right, the technology's been out here, blah, blah, blah, right? You know, you can make the case for, you know, Sonnet 4.5, which may be a superior coding model. We'll see how the benchmarks come out with GPT5 Pro, GPT5 codecs, etc. But I think ultimately the users, that's where the long term gets played out. And OpenAI comes with the users. So when OpenAI creates something new, a new innovation, a new technology, they're actually the ones pushing the envelope here. And the envelope has been pushed, I think, with Codex and its impact on traditional software development. Number two, ChatGPT wants to be your operating system and this might be the death of the web. We've already talked about this. Not just with the AI answers engines. Deep research, right? I use deep research now. So much more where I used to just have to go to 20, 30, 40 websites to do my own research. But now with apps, I mean, we are bringing the entire app experience inside ChatGPT, right? We didn't hear any, you know, any new news on their reported forthcoming agentic browser, right? But they're literally bringing entire website experiences. Look at Coursera, look at Zillow, right? They're literally demoing, oh, a Coursera video playing and then ChatGPT takes notes on it for you. And you can talk with ChatGPT about the video that's being played inside of the Coursera app in Chat gbt in Zillow, right? You can search Zillow and search with natural language. Oh, find me something in the western Chicago suburbs with a pool that's, you know, less than $800,000 and within a half mile walk to a park, right? You can literally just with natural language, interact with Zillow without ever having to go to Sillo. Same thing. Canva. You can create and edit and iterate slides and designs in Canva in Canva, much different than their Canva GPT or even the Canva connector. The Canva app brings the experience inside of ChatGPT. But I think this is also the beginning of the end of, well, the death of using the web. OpenAI wants to be your operating system, so they're bringing as much as possible into their platform. Why? Well, they're rolling out this whole checkout feature thing, right? Again, going the Amazon route, but more importantly the Apple route. Which brings us to number one. The number one thing I think people overlooked is the real. And I think it became very apparent after Dev Day. In OpenAI's mind, I don't think they're going after Google or Microsoft. I think they're going after Apple, right? Because look, Apple has dominated on the software side. The software side led to their hardware sales. If you look at the history of Apple and that created ultimately users, which created ecosystem lock inside that is the same playbook that OpenAI is taking. They are crushing everyone else, even Google. Even though Google is taking shares back from OpenAI and others. I Don't think anyone will be able to catch OpenAI anytime soon if the current pace keeps up unless something drastic changes. And this we've, we've talked about on the show before, you know, OpenAI has reportedly hired dozens of hardware engineers from Apple. The, the partnership with Apple designer Johnny Ives, the direct relationship with suppliers. They're going after AI hardware, but they're first doing it by winning the quote unquote software game and winning the user games. And I think the end goal here, and we're starting to see it with the apps in creating the ecosystem lock in. So when you bring the apps inside ChatGPT, but then you also bring the payment gateway, the agentic payment gateway, and being able to check out and buy things, you know, from, from Etsy or Shopify, which that's what's live now, into Chat GPT. Not only are they creating lock in and trying to kill off the traditional web, well, they're going specifically after Apple. So maybe Sam Altman is Steve Jobs too, or that's what he's going for. Although I really think he should have wore the double popped collar polo on dev day. All right, so that's a wrap, y'. All. All right. We didn't get to 25 minutes, but not too bad, right? So if you missed dev day, we just wrapped it all up for you. We talked about what I think the two big things are ChatGPT's new agent builder. So even if you are not a technical in, you can go in there in a drag and drop canvas and start to build an agent that you can deploy on your website. ChatGPT bringing actual apps in the UI, the user interface of different apps into ChatGPT. And then I went over five pretty big takeaways that I think most people miss. And hey, podcast people, can you do something for me if you're listening on Spotify, I'm actually trying something. I'm putting a poll in there, putting a poll on our Spotify. So scroll to the bottom of the show notes, all right. And let me know, do you want to see an episode maybe later this week going over in more depth. Do you want apps so ChatGPT apps or the ChatGPT agent builder? So go check out that poll. Let me know also live stream audience, if you found this helpful, please repost this. Got a little something extra for you. I actually had three more hot takes that I knew I wouldn't have time for. So if you do repost this show, I will send those over to you. They are ready to go. Couldn't fit them all. So if you miss anything, please go to your everydayai.com, sign up for the free daily newsletter. We're going to be recapping today's show and giving you everything else that you need to know. Thank you for tuning in. 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.