Resources:
Join the discussion on LinkedIn: Got something to say? Let us know on LinkedIn and network with other AI leaders
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson: LinkedIn Profile
Start Here Series in our Inner Circle Community: Join for free access
AI Productivity Tools: Critical Insights from Microsoft Build, OpenAI Codex, and GitHub Copilot Advancements
Major technology providers unveiled several actionable updates impacting how organizations can deploy AI tools to optimize knowledge work. From OpenAI's new role-specific Codex plugins to always-on Copilot agents and seamless enterprise integration across AWS and Microsoft platforms, there are specific takeaways for business leaders managing teams, workflows, and technology procurement.
OpenAI Codex Plugins: Role-Specific AI for Enterprise Workflows
OpenAI introduced six targeted Codex plugins, designed for roles such as sales, data analytics, product design, creative production, investment banking, and public equity investing. Each plugin bundles relevant app integrations (62 enterprise apps including Snowflake, Figma, Salesforce, Databricks) and skills (over 110 automated abilities), enabling professionals to supercharge tasks like sales opportunity prep, financial research, and prototyping without custom prompt engineering or manual workflows.
Critical to enterprise adoption, access control sits with workspace administrators, ensuring app permissions align with organizational governance. These plugins transition AI from developer-centric tooling to accessible, role-aware agents for analysts, marketers, product teams, and consultants. For organizations using ChatGPT Team, Business, or Enterprise, Codex plugins are immediately accessible, significantly accelerating the deployment of AI-driven process automation and knowledge work orchestration.
Microsoft AI Image Generation: Enhanced Integration for PowerPoint and OneDrive
Microsoft announced a major upgrade to its in-house image generation models, bringing top-tier capabilities into everyday productivity platforms. The new image model—with a Flash variant for developers—has been integrated directly into PowerPoint, reducing friction in content workflows. This model delivers improvements in text rendering, stylized illustrations, and commercial image generation, and is also rolling out to OneDrive.
In practical terms, organizations using Microsoft 365 can now create, edit, and deploy images without juggling third-party tools or external services, preserving compliance and security by staying within Microsoft’s ecosystem. This is especially critical for enterprises with strict IT controls, making AI-powered visual asset creation fully accessible in locked-down environments.
Canva and Codex: Context-Aware AI Design Workflows
Canva now functions as a plugin within Codex, with the Canva app in ChatGPT receiving a substantial design upgrade. Users can create and edit designs within the ChatGPT interface, enjoying full-screen preview and interactive editing that closely mirrors the native Canva experience. This integration means automated reports, marketing materials, and internal presentations are generated in context—reducing copy-paste friction and retaining continuity across AI-assisted content creation.
For development teams, Canva’s MCP server update offers further extensibility, and rollout covers free, Plus, and Pro plan users outside the EU. The direct linkage between KPIs, data workflows, and design assets means teams can automate not just information gathering but also the production of visually-rich deliverables, removing traditional barriers between analytics and creative execution.
GitHub Copilot App: Agent-Native Command Center for Autonomous Work
Microsoft’s GitHub Copilot App—introduced at the Build conference—marks a pivotal shift to agent-native desktop experiences. The app serves as a centralized command center outside the developer IDE, featuring a ‘My Work’ view that aggregates active sessions, issues, pull requests, and background automations. Notably, the Canvas feature provides real-time visibility and verification of agent-led work, ensuring transparency and manageability in autonomous workflows.
GitHub Copilot App supports multiple models (including offerings from Microsoft, OpenAI, and Anthropic) and is in preview for Pro, Plus, Max Business, and Enterprise customers on Windows, Mac, and Linux. This tool allows enterprises to secure, manage, and track parallel AI agents, bringing clarity and operational control to complex software projects, while consolidating automation management in a single native interface.
Microsoft Autopilot Agents: Proactive Autonomous Workflow Across Microsoft 365
Microsoft introduced its new ‘Autopilot’ category, with ‘Scout’ as the first example of always-on autonomous agents designed for enterprise environments. Scout integrates across Microsoft 365 apps—Teams, Outlook, OneDrive, SharePoint—proactively handling tasks such as meeting preparation, conflict resolution, and routine scheduling.
Notably, every agent action is governed by enterprise-grade identity (Intra ID), enabling managed, auditable interactions aligned with corporate security protocols. Instead of reaction-based automation, the system continuously monitors workflows and acts without user intervention, while IT retains oversight and traceability for all operations.
AWS Bedrock: OpenAI Integration for Enterprise Governance
Amazon Bedrock now includes general availability for OpenAI’s GPT-4, GPT-4.5, GPT-4.5 High, and Codex coding agents. This integration enables AWS customers to access leading language models within their existing security, billing, and procurement frameworks, removing a significant adoption barrier for large organizations.
The inclusion is particularly valuable given OpenAI's reputation for token efficiency—providing comparable intelligence for 30% to 50% less cost than certain competitors like Anthropic's Opus 4.8. For organizations scaling AI workloads and seeking to optimize cloud spending, Bedrock’s support for OpenAI models offers a pragmatic pathway to balance cost, capability, and compliance.
Codex Sites: Rapid Enterprise App Deployment and Collaboration
OpenAI’s ‘Codex Sites’ feature allows users to generate and deploy functional web apps—such as dashboards, knowledge bases, and internal tools—from a single prompt. Each site comes with dynamic internal data connectivity, secure sign-in via ChatGPT, and enterprise-compliant access controls.
For teams needing interactive dashboards or project management tools, Codex Sites eliminates the engineering overhead and security hurdles of traditional app development. As internal data is updated, so are the linked apps, ensuring accuracy and iteration speed. This addition addresses the gap between ad-hoc analysis and enterprise-grade tooling, offering scalable, shareable, and secure options for cross-team collaboration without custom development cycles.
Conclusion: Coordinated AI Progress Unlocks Immediate Productivity Value
The integration of context-aware AI tools into business operations—from the specificity of role-based plugins and design workflows to proactive agents and enterprise cloud compatibility—creates actionable pathways to enhance internal productivity, maintain compliance, and reduce manual overhead. These updates offer decision-makers precise avenues for implementing AI, moving beyond experimental pilots to tangible workflow transformation, cost optimization, and risk management in modern enterprise environments.
Topics Covered in This Episode:
- OpenAI Codex Role-Specific Plugins Launch
- Microsoft Build Conference AI Feature Releases
- ChatGPT Memory and Business Account Upgrades
- Microsoft Flash Image Model for PowerPoint
- Canva Integrated with ChatGPT and Codex
- GitHub Copilot Standalone Desktop App Preview
- Microsoft Autopilot Always-On Work Agents
- OpenAI Models Now Available on AWS Bedrock
- Codex Sites: AI-Built Internal Web Apps
Episode Transcript
Jordan Wilson [00:00:17]:
Normally, when one big tech company takes center stage, some of its competitors actually sneak backstage and steal the thunder. That may have been the case this week as Microsoft rolled out a handful of useful AI updates at its build conference. It was actually OpenAI that probably made a bigger splash with some pretty big money moves when it came to codex and chat g p t, and specifically, how they work with business accounts. Don't worry. We still have some other AI updates, like if you're a Canva user, but even those kind of involve chat g p t and codecs. And I know you don't have an extra five hours every single day to keep up with all of these new AI updates, but I do. And that's what Fridays are for as we bring you AI features on Fridays. So your one stop shop, if you just wanna know what can I use today to help grow my company and my career, that's exactly what we do on this Friday show? So on today's show, you're going to learn the two big moves, from OpenAI that could shut down some anthropic advantages.
Jordan Wilson [00:01:33]:
You're gonna see how Microsoft is planning always on agents, and you're gonna see how AWS actually got way more useful for AI users. Alright. Let's get into it. Welcome to Everyday AI. What's going on y'all? I'm Jordan, and this is for you. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me keep up with the nonstop AI updates. Yeah. There's dozens every week.
Jordan Wilson [00:02:00]:
I tell you what's important, what's BS. I show you how to use it, and then you take the credit. You get to be the smartest person in AI and grow your company and career. So it starts here with the unedited, unscripted livestream podcast, but make sure you go to our website, youreverydayai.com. It's a cheat code on there. There's literally thousands of hours of free content, but make sure you go drop your email in there so you can go keep up with the highlights from today's show as well as all the other AI news you need to know to stay ahead. Alright. Let's get straight to it.
Jordan Wilson [00:02:32]:
A ton of big updates this week. So, generally, it seems like five is kind of the, kind of the magic number, in terms of how many updates we'd like to go over this week. And, unfortunately, there was a lot of really good ones that I was kind of excited to talk about that didn't make the cut. I mean, Gemma four, they have a new 12 b model. Nemo Tron Ultra three. Chat GPT got a new memory upgrade. You know, Windows users got codecs, computer use, and mobile access. There's a new Microsoft Flash model.
Jordan Wilson [00:03:06]:
Right? So many of these things did not make our top seven mainly because Microsoft had their build conference and announced a ton of new AI features you could use today, and chat GPT was shipping everything. We had a quiet week from anthropic, but I expect that to change probably next week. Alright. Let's start off with our first big AI update, and that is making money moves. Yeah. The open AI brass got all dressed up and fancy and held a, essentially, hey, enterprise leaders come, and we're gonna show you how good codecs is. And the way that they've led with this is with plug ins. So here's what's new.
Jordan Wilson [00:03:47]:
Now you have six role specific plugins for codecs. So there's a sales data and now, data analytics, product design, creative production, investment banking, and public equity investing. So each bundle, each bundles role skills, app integrations, starter packs, and workflow guidance. So collectively, so altogether, they package 62 different apps together. So, you know, the big names like Snowflake, Figma, Salesforce, Databricks, Hacks Clay. Right? So the enterprise apps that people are using, and then a 110 automated skills. So if you don't know what a plug in is inside of, codecs, RIP old school plug ins inside of GPT four, but essentially plug ins inside of codecs, they package together apps so it can grab data from, dynamic data from the apps that your team uses every day, and then it stacks on type of its skills. So sometimes, you know, some of these, plugins they announced might take advantage of, you know, three different apps and three different skills and kind of bundle them all together.
Jordan Wilson [00:04:57]:
So here's who has access. It's live. So if you have a team chat g p t account, that means you have access to it. So, obviously, to use these because they're codex, you will have to download codex. So, you know, if you are a Chatt GPT enterprise organization or a Chatt GPT business organization, well, any Chatt GPT subscription means you have codex as well. The usage is kind of shared between the two. So, also, if you're on a, plus, account or a pro account, you do have access to these as well. Also, what's important to know because these are very powerful, workspace admins control the underlying app permissions.
Jordan Wilson [00:05:37]:
So if you are on a large larger organization, yeah, you're not gonna be able to just go in there and start tinkering around with these things. You're gonna have to get your workspace admin to turn them on. So here's why it's useful. Well, it turns codecs from a coding tool into a role aware work agent. So non developers can run sales prep, dashboards, prototypes, and financial research without even building from scratch. And this is pretty big because, you know, essentially, where these plugins got you, you know, for a power user like me, this isn't a huge step in function. Right? I'd say for everyone else, this is an an enormous step. So this is essentially what I was doing.
Jordan Wilson [00:06:17]:
Right? I would, you know, go and grab, you know, two or three different apps that really worked in conjunction with each other. I would either write a very long prompt, I would dictate one, or I would use one or two saved skills that I already had. So that's kind of what how power users have been taking advantage of codex, and everyone's, like, losing their mind, you know, over the last, like, month because they're figuring out what I've been saying since February. Like, codex is unfair. It's so far ahead of everything else. Now this helps you realize this by kind of giving you those starter packs out of the gate that package together these, you know, enterprise apps and skills. So who's gonna find this valuable? I mean, if you're a heavy chat GPT team, you. But also analysts, marketers, operators, product designers, researchers, investors, bankers.
Jordan Wilson [00:07:07]:
Right? Some of those, you know, roles that maybe three years ago, these companies, these type of individuals or these type of enterprises were really touching generative AI. Right? Because three years ago, you know, models didn't even know two plus two. Right? Now they're winning IMO gold medals. Right? The International Math Olympiad. So it's it's gonna be, I think, a rush now, as these, you know, financial sector consultancy sectors are gonna start ushering in both from, you know, Claude, which really popularized, kind of the skills. And now with codex and what we got, here, with these plugins. So here is, give you the little spiel from OpenAI, what they said. They said make codex work the way your team does.
Jordan Wilson [00:07:54]:
Codecs is most powerful when it works the way your team does, connected to the tools you use, and ready to create the materials you need. Plugins help codecs work with tools, context, and workflows your team already uses. Today, we're launching six new role specific plug ins that make codex useful for more kinds of knowledge work, no coding required. And then they do say, each role specific plug in bundles the relevant apps, skills, instructions, and workflows. Together, they include 62 popular apps and a 110 skills. So then, yeah, they have the data analytics plugin, the creative production plugin, the sales plugin, product design plugin, public equity investing plugin, and the investment banking plugin. Alright. So, make sure you check out today's newsletter.
Jordan Wilson [00:08:37]:
They do have some nice little demo, demo videos, showing those new plug ins in action. Alright. Let's go next. So finally, even though this one is kind of not new, it is also kind of new. Let me explain what I mean. So next, we have Microsoft's new image model, which looks really good. Alright. So when they did announce this, it was like a top two or three model.
Jordan Wilson [00:09:02]:
It kinda got knocked out a couple of slots because there were some nice open source, models that were released. Well, there were some open source ones and there were some, you know, closed source, ideogram four came out as an example. So kinda got bumped down the rankings, but regardless, Microsoft is now essentially a top three company when it comes to AI image models. So here's why it's kind of new, but also kind of not new. So you were able to use this previously on arena, but that was the only spot. Right? So now for the first time, Microsoft is making it publicly available. So this is an upgraded in house Microsoft image generation model with a new Flash variant for developers that they announced at the build conference. Also, FYI, I did do a whole rundown on the build conference, so I'm not gonna go over this, the Microsoft ones too much.
Jordan Wilson [00:09:54]:
So make sure you go listen to yesterday's episode. That's episode seven ninety one. Alright. So, this is their new model. Microsoft claims improvements in text rendering, stylized illustrations, and commercial imagery benchmark near the top of LM arena image rankings. So this is live. It is out. So it's already integrated, which I like that they're doing this into PowerPoint, and it's currently rolling out to OneDrive as well.
Jordan Wilson [00:10:21]:
Developers can access it through Microsoft Foundry, and then the Flash variant that, developers use is there as well. So here's why it's useful. Well, it lets you generate and edit images directly inside of PowerPoint. So you kind of remove that round trip that you had to do before, you know, of using something like, nano banana using, like, chat g b t images. Right? Or I think this opens up a whole new world of possibility. So if your organization is, like, I don't know, like, 80% of enterprises out there and locks down tools and pretty much you're using Microsoft Copilot only, right, well, now you actually have a powerful frontier level image generator. Right? So I have a couple, examples here, on my screen. And I mean, they're really good.
Jordan Wilson [00:11:09]:
Right? Again, I know a lot of people, and I talked about this yesterday, are knocking on Microsoft because they're not number one at anything, but I don't think they need to be. Right? I think they can be second or third best at, like, everything, and still remain very relevant for decades to come at least when it comes to the AI race. Alright. Let's go next. So next, this one is small, but I'm mentioning this one because I think the implications are big, especially for small business owners, marketers, etcetera. So here's what's new. Canva is now available as a plugin in codecs, and their, the Canva app in chat g p t got a full design creation upgrade. So now you have editing creation in full screen preview inside of chat g b t.
Jordan Wilson [00:12:01]:
So we have the introduction of Canva into codecs, and we have essentially a read and editing version of, chat g b t inside of Canva. So the old Canva app in chat g b t left a lot to be desired. I had demoed it on the show multiple times. It didn't really work as advertised. The newer version, a little bit better. Right? So you actually have a much more, true to interface, true to Canva editing and creation experience inside of chat GPT. Right before, you you know, the the preview thumbnail didn't always work. You couldn't flip through the pages.
Jordan Wilson [00:12:37]:
You really couldn't do a whole lot of normal Canva editing. Now you can. So, developers also can build around this with the Canva MCP server, which is updated as well. So here's who has access. So if you have the Canva app in chat g p t, well, now you have full design capabilities. So it's rolling out now to ChatGPT users on the free plus and pro plans, not in counting the not counting the EU. And then the Canva plug in in codecs is available to all codecs users via the plug in directory. Alright.
Jordan Wilson [00:13:12]:
So here's why it's useful. It's well, now the design layer travels with the work. Right? That's huge. You know, it it it cuts down on the copy and pasting, but also just how codecs and chat g p t can carry context right throughout the course of a conversation. So let's say that you're, you you know, updating, you know, your KPIs. You update them monthly. Right? And you're going back and forth in chat g p t or codex, grabbing all this information from your other dynamic data sources. That at the end, well, you can just say, put this into a one pager on Canva, and it's gonna build it right there in the app, and you can iterate with it.
Jordan Wilson [00:13:51]:
So instead of having to go through and be like, oh my gosh. Now I have to, you know, summarize and start pulling out bullet points through all this. That's the great thing about the agentic. Right? Ready? For if you didn't go listen to this, if you didn't listen to this one, make sure you go listen to our agentic. I'm actually gonna look it up here. So that's our agentic context, Carrie, episode in the start here series. What one was that? That was episode seven sixty two seven sixty two. Go listen to that if you wanna understand agentic context carry.
Jordan Wilson [00:14:23]:
But that's what makes this new editing, functionality inside chat g p t within codecs, or sorry, Canva inside of chat g p t and codecs really powerful. So here's what Canva says. It says the new plugin for codecs complements the much loved Canva app in Chat GPT, which unlocks design creation and editing from inside the chat. That means you can generate context aware designs, translate assets, edit text, and more, all simply by asking chat g p t. Enter a prompt, and a new Canva design will appear ready to go. The Canva app in chat g p t delivers the visual experience in an AI chat, delivers the most visual experience in an AI chat yet featuring full screen design preview directly within the main interface. For the first time, you can kick start designs based on your chat, refine them, and take them into Canva without losing creative momentum. And I think the last, sentence there, that's the whole crux of this.
Jordan Wilson [00:15:21]:
Right? Without losing the creative momentum, which is funny to say. Right? Because that's Canva implying that, oh, well, for the creative momentum, you can just do it inside chat g b t. Right? I've been talking about this now for a year. Right? Way before the headless, you know, movements. I think it it was actually around some of these more, you know, visual interfaces when ChatGPT introduced apps in 2025. You know, I said they're bringing in interfaces. Right? So the normal interfacing that you would do with a website, you can now do a lot of those basic things within ChatGPT. Right? Using actual features and functions, using the the the gooey of, you know, some of these pieces of software and apps that you would use on an ongoing basis, it's now all there for you.
Jordan Wilson [00:16:11]:
Alright. Let's keep going more Microsoft. We have a new GitHub Copilot app. Yes. Microsoft GitHub, is in on the super app game. This is the precursor to the true Copilot super app. But right now, this is the new standalone desktop app released at Build that acts as an agent native command center for managing AI agents outside of the IDE. So inside the new app, there's a new my work view that shows active sessions, issues, pull requests, and background automations across connected repos, and each session holds its own isolated Git work tree for parallel agents.
Jordan Wilson [00:16:56]:
So this new update also adds a canvas, so a space to make agent work visible and verifiable. So this is something, you know, at the build conference that new canvas element inside GitHub Copilot, was actually, I think, one of the highlights of the entire two and a half hour keynote. Right? I think that was one of the features that got a lot of ooze and awe from the crowd. So, you know, pretty cool to see that within, the desktop elements. So, you know, to to summarize this. Right? So, for our our audience, if you haven't seen it, this is for all intent purposes, it looks codecs codecs esque. Right? I'm not gonna say it it looks as codecs as Gemini's anti gravity two point o app, which looks like a clone, but it does kind of just look like codex, which I'm guessing this just means that's kind of the future, right, in interface that looks much like the codex interface. Right? So you have your work trees, you have your chats on the left hand side, you have your project folders, you have your automations on the upper left, right, your settings on the lower left.
Jordan Wilson [00:18:07]:
So if you've used codex or if maybe you use codex personally, but you can't use it at work, well, I think the GitHub Copilot app could be a, a great option for you because, you know, some of the same capabilities, obviously, you know, actually an advantage to GitHub Copilot is you can use a variety of models. So you could use Microsoft's new models, which are, you know, top 10 ish, give or take. Right? But you can also use OpenAI models and anthropic models. So there's actually some cool, advantages there. Like I said, that and the new Canvas feature, pretty cool. So, here's who has access right now. So it's in preview, but it is open to a lot of customers. So, also, if you have existing GitHub Copilot Pro, Pro Plus, Max Business, or Enterprise, customers, than you are in.
Jordan Wilson [00:19:01]:
So you just have to make sure to download, this for Windows 11, or there's a Mac version and a Linux version as well. Actually, I spent a disproportionate amount of my time at Microsoft build, kind of watching some of these GitHub Copilot demos, talking to some of the team building the product. So, looks very good. Right? The new updated version. So here is what Microsoft says. They says, they say, today, we're introducing can't, canvases. Oh, that's the canvas part. Let me read the whole part.
Jordan Wilson [00:19:36]:
Here we go. So they say, where where did it go? Where did it go? There we go. Alright. You start the day with three pieces of work already in motion. One agent is investigating a production bug. Another is implementing a back backlog issue. A third is working through a review feedback on a pull request. Each is running its own isolated environment, producing changes you can inspect, redirect, test, and manage.
Jordan Wilson [00:20:06]:
You need an environment that can keep up. The new GitHub Copilot app is the agent native desktop experience built on GitHub. From a single my work view, you can see work in motion across connected, repos, active sessions, issues, pull requests, and background automations. Copilot app is now available in technical preview for existing Copilot plus, sorry, Copilot Pro Pro Plus business and enterprise users. Yeah. So, the positives, it looks really good, really clean, new features. People are loving that aspect. The downside, people are absolutely hating hating the new GitHub usage.
Jordan Wilson [00:20:46]:
Alright. So, it's been, a little bit of an uproar online. Right? Essentially, all the big companies except, OpenAI are starting to subsidize tokens way less or kind of stopping to subsidize one via subscriptions. Right? There's been this, I wouldn't even say it's an elephant in the room. It's become an expectation, right, that within your, subscription, you know, you're usually getting 10 to 50 x, in terms of actual token usage, what you'd be getting then if you paid, you know, a la carte via the API. So a lot of the companies that started with Anthropic, and Tropic's always been the stingiest. Google after their IO conference two weeks ago, they started to implement pretty strict usage limits on their $20 a month plans. And the same thing with the GitHub plans, people are actually pretty upset about it online, so it'll be interesting to see what Microsoft does about that.
Jordan Wilson [00:21:40]:
Alright. Let's keep moving. We have three more big ones. This one, probably the biggest, headline grabber. We'll see if this turns out to be anything of substance in the long run, but Microsoft announced their autopilot. So this is a kind of a new breed of copilot. So they announced this at build, and it's their always on autonomous work agents built on OpenClaw and the new work IQ with its own govern intra identity. So you can do the whole observability and trust, you know, thing with your agents.
Jordan Wilson [00:22:15]:
And the cool thing is the autopilots can run across teams, Outlook, OneDrive, SharePoint, and local device actions, proactively handling meeting prep, scheduling conflicts, and routine tasks without being asked each chime. So the first one that Microsoft introduced is called Scout. So Microsoft said, we are also introducing Microsoft Scout, our first autopilot agent. So autopilot is the category, and Scout is the first one that is available. So it says Microsoft Scout is integrated across the Microsoft three sixty five apps you use every day, keeping it grounded in your flow of work. It operates across cloud, desktop, and web connected to Teams, Outlook, OneDrive, and SharePoint, and to the data that powers your day, including chats, email, calendar, and contacts. You interact, you interact with it in Teams and extends its reach through the desktop app to your browser, local files, and model context protocol servers. Alright.
Jordan Wilson [00:23:17]:
So here's who has access. So it is being rolled out now to Frontier customers. I do see some people have it, but it looks like a slower rollout, and it is tied to Microsoft's enterprise agent stack. So here's why it's useful. Well, it shifts agents from kinda just being ask and wait to being proactive. Right? So it'll watch your different work surfaces, and then acts on routines, on on routine items that you can, then check its work on. The other good thing, well, there's the governed intra ID, which means that IT can manage and audit it like any other corporate identity. Right? So this isn't, you know, Scout is not something that's gonna be, you know, accidentally, you know, deleting your CRM and, you know, emailing customers new prices.
Jordan Wilson [00:24:03]:
Right? Because it's governed, and there's, the intra ID that you can go back and check all of its traces just like you could a normal employee. Alright. Two more big ones. First, a lot more companies are gonna start using ChatGPT. This is one I wasn't even gonna mention it, but I noticed that it got, like, no media play. So you probably don't know this. So that's why unless you're reading our daily newsletter because it's a big deal, and we did feature in our AI news. But I think people are overlooking what this means, not just for OpenAI and ChatGPT, but the overall landscape.
Jordan Wilson [00:24:42]:
So this is new. AWS is now supporting OpenAI's models inside of Amazon Bedrock. So that includes OpenAI's GPT four, five four, five five, and the codex coding agents are all generally available on Amazon Bedrock. So that's big. The GPT models are called via the response API on Bedrock's inference engine, and codex runs through the codex app, the command line interface, and the IDE integrations in Versus Code, JetBrains, Xcode, etcetera, with inference routed through bedrock. So this is big because right now, any AWS customer can now be using the GPT models. So this is, if you're an AWS customer and for whatever reason, which I think is what a lot of companies are, they're coming to their token reckoning. Right? It's actually one of the more interesting conversations I heard a lot out at Microsoft build.
Jordan Wilson [00:25:48]:
You know, people were talking about using small language models to cut back on your Claude subscriptions. Right? Use this new, you know, Microsoft model to cut back on your Claude, you know, API bill. Right? I I I think that there's a a there's been a whiplash, to anthropics models. I have a lot of takes on that. One of them is the they're the most token inefficient models on the market. You know, they have to use more tokens than anyone else to get the same level of intelligence. Also, the downtime, they run-in loops. So Interopix models, although good on the surface, they are grossly expensive for what you actually get in terms of intelligence.
Jordan Wilson [00:26:30]:
So I know a lot of big enterprises on AWS are gonna like this because right, at least right now, OpenAI's models are much more token efficient. And when you look at model to model, it's pretty much a coin flip. But, you you you know, depending on what, index you look at, but third party indexes show that OpenAI, their g p t five five, high gets the same level of intelligence for about 30 ish percent, cheap 30 to 50% cheaper, than, the Opus 4.8. So pretty big news there, and here's why it's useful. Well, it lets enterprises run Frontier OpenAI models inside their existing AWS security governance, procurement, and billing. So removes a major barrier to production deployment. Yeah. I mean, I remember talking to companies a year ago, 18 ago, you know, talking about this.
Jordan Wilson [00:27:28]:
Right? Because, you know, it started with, you you know, obviously, having access to Microsoft's models, or sorry, not Microsoft's models, Amazon's models, right, which I don't think a lot of people like. Then they got Anthropics models, and I think people were happy. And, you know, well, once you start using other models, you know, if you're upgrading from the AWS models to anthropic models and you like that, but you're like, man, this is a lot of money. Well, now you have a good option. Alright. Let's move on to our last one. And this one, you probably missed it. You probably missed it.
Jordan Wilson [00:28:06]:
I didn't, I'm in loving this. So codex sites. All right, before we get into the details, let me just say this. I think this could be the future. You're right. Maybe not codec sites. Maybe this concept of how teams communicate together in an AI native workspace. Alright.
Jordan Wilson [00:28:25]:
It's that big potentially if it works. Right? This is one of those things if OpenAI keeps developing it. I mean, we'll see. You know, luckily, they've continued to develop to develop workspace agents, but some projects, you know, get one or two updates, and then they kinda get abandoned. So I really hope sites is something that sticks around because I think in terms of a first version of something that has the potential to really change, how we work, this has to be a top five release from OpenAI ever. In terms of going from zero to enterprise potential, this one's big. So sites, codecs sites is a new feature that builds, deploys, and hosts interactive web apps from just standard prompts, returning a shareable URL. So this handles dashboards, project trackers, internal tools, knowledge bases, even games as full stack JavaScript type scrap apps.
Jordan Wilson [00:29:23]:
So it includes a sign in with chat g p t access in workspace internal data and file storage. So this is right now in preview, but it is enabled by default for Chad GPT business workspaces in Chad GPT enterprise teams. So right now, to get access, you look in your code app, codex app sidebar, and then click the at button and then install, first you'll have to, install it and then you're gonna see these sites plugin. So this is hugely useful. So this is essentially lovable for your team. Right? If you've heard of the lovable's and the bolt new, right, replit, all of these big kind of companies that are, you know, kind of like full stack apps. That's what this is. So this collapses the whole, like, describe an app, get a hosted shareable app into one step.
Jordan Wilson [00:30:24]:
But it makes it instantly shareable, and secure across your team. So the fact that now you can all of your enterprise data, you can connect to it dynamically. You can use the new sites plugin, spin up an app, share that with your team. It is secure enterprise, you know, enterprise security, and it's instantly updated and shareable. Right? So this was a big thing that was, I think a big gap in the market. As an example, I used to love ChatChippy Tea Canvas, but, you you guys notice when I do these things where I spend a lot of time and I build these nice tools and directories for you guys, I generally would share them in Gemini Canvas. That's because with Chad GbT Canvas, it was kind of harder to share those things. So number one, this makes it a little bit easier.
Jordan Wilson [00:31:18]:
But number two, this is dynamically updated within your organization. So as you get fresh data rolling in, just like a database, it's going to update. Right? So the app is going to update as your data updates, and then you can obviously iterate this on this or work on this with your team. So it keeps the internal tools behind workspace, workplace off, so teams can share these apps without exposing data publicly. And this inherits whatever chat GBT or enterprise security settings that you have. Alright? So no training on your data by default if you're a business or enterprise, company. And so here's, I think, who's gonna find it valuable. So, I mean, operators and team leaders who need a quick internal dashboard or tracker without pulling in engineering.
Jordan Wilson [00:32:07]:
So, you know, if if you've had something and you're like, oh, we can vibe code this, but you're like, that it's we need a little bit more security. How are we gonna share this? How are we gonna keep it updated? You know, what about from a database perspective? Well, this kind of answers a lot of those questions. So it'll be interesting, to see how the market reacts to this. Right? I'm looking at, you know, the smaller project management tools. Right? Are they gonna start to get used less? Right? Especially when the original data connectors are headlessly available within codecs and CHEGT. Right? So this just kind of connects. And and this is for me, I'm so excited about this one because this this is one of the big things I was using Claude code for. I was using codex for, right, essentially just building myself these interactive dashboards, but now it's well, it's a plug in that's built by the experts at OpenAI.
Jordan Wilson [00:33:05]:
So pretty big. Here's what OpenAI said about it. They said sites are a new kind of canvas for your ideas. Codecs can take your ideas, analysis, and plans, and turn them into dashboards, planners, review workspaces, project boards, galleries, and lightweight tools. Today, sites can be shared with anyone in your workspace via URL, giving teams a shared place to explore work, contribute input, track progress, and make decisions together. Then they say at the bottom, instead of adapting work to the limits of a single tool or file, teams can create sites that fit the work, and sites aren't static. They can also help track progress for a major product, help guide customer service reps, or act as a repository for your team's creative briefs. So, yeah, make sure you check out today's newsletter, and I'll send you the link so you can kind of check out, some of these demos that they did.
Jordan Wilson [00:34:00]:
Really cool. Alright. So that is a wrap for our Friday features. A quick rundown of seven AI tools you can use today. Yes. All of these things I talked about, they're live. They're not coming later. These aren't just random news pieces.
Jordan Wilson [00:34:16]:
Right? On Monday, we go over the AI news so you can understand, you know, kind of what's happened, what's happening in the business world, what's happening, happening, on the politics side, the drama of the big companies, the rumors. Right? But on Fridays, we put AI to work for you by showing you all the new AI features that you can use today. I hope this was helpful. So helpful. If so, tell someone about it. Alright? Share this if this been helpful for you. If you can subscribe to the podcast, I would really appreciate that. And then make sure you go to your everydayai.com.
Jordan Wilson [00:34:53]:
Go sign up for the free daily newsletter. We're gonna be recapping these seven, highlights from today's show. Send you the link to all those so you can check them out and a whole lot more. Thank you for tuning in. We hope to see you next week and every day for more everyday AI. Thanks, y'all.