Everyday AI Podcast – An AI and ChatGPT Podcast - Ep 724: Trump bans Anthropic, OpenAI signs Pentagon deal, big AI goes agentic and more AI news

Episode Date: March 2, 2026

While Anthropic and the Pentagon fought, OpenAI swooped in to secure a big deal. But at what cost? And while it seemed like the entire AI news world was wrapped up in the Anthropic-Trump-OpenAI drama, the rest of big tech went nuts. Microsoft teased something agentic, Claude actually shipped it, and Perplexity dropped probably its most important product to date. This week’s theme apparently: drama and agents. We’ll get you caught up on all of the AI News That Matters. 👇Trump bans Anthropic, OpenAI signs Pentagon deal, big AI goes agentic and more AI news -- An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:OpenAI Pentagon Deal Security ProtocolsAnthropic Trump Ban and Government ResponseAnthropic Supply Chain Risk Legal ChallengeGoogle Nano Banana 2 Image Model LaunchNano Banana 2 Benchmark and Feature UpgradesOpenAI Enterprise AI Frontier Consulting PartnershipsAnthropic Model Distillation by Chinese AI LabsPerplexity Computer Cloud-Based Agentic AI ReleaseMicrosoft Copilot Tasks Agentic AI AnnouncementAnthropic Claude CoWork Scheduled Task AutomationTimestamps:00:00 "OpenAI-Pentagon Deal Sparks Debate"03:41 OpenAI's AI Safety Agreement08:25 Anthropic vs. U.S. Government Tensions12:54 "Nano Banana 2 Revolutionizes AI"15:38 "Revolutionizing AI for Visuals"19:27 "AI Outperforming Skilled Professionals"21:33 Anthropic Accuses China of AI Theft26:27 "AI Supremacy and Global Impact"29:24 AI Automates Tasks and Emails33:31 "Perplexity Unveils Agentic AI Tool"37:00 Perplexity's Evolution: Computer Spotlight41:10 Trump Meets Tech Leaders Over AI42:41 "AI Updates Made Simple"Keywords: Trump AI ban, Anthropic, OpenAI Pentagon deal, Department of Defense AI, AI military use, ethical boundaries in AI, federal AI regulations, Sam Altman, AI contract enforcement,Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

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. If you happen to be on vacation this past week and you came back and you're like,

what the heck happened in the AI? Well, you're not alone because even if you were following what happened in the AI world every single day for the past week, you're probably also wondering what the heck happened. I mean, Open AI inked a deal with the Pentagon when it didn't seem like that was going to be possible after Anthropic and President Trump and the Department of War were bickering over AI basically all week. And then you got nanobanana 2 come out. That's the best AI image model in the world.

And all of big tech, well, they just decided that they were going to release actually useful agentic AI all at the same time. So yeah, if you're feeling a little behind on. What happened in AI this week. You're not alone. Jump on the bus. I'm driving. All right.

Let's get into it. If you're new here, welcome to Everyday AI. My name is Jordan Wilson. And this thing's for you. Everyday AI is your daily live stream podcast and free daily newsletter, helping everyday business leaders like you and me keep up and get ahead with everything that's happening in the AI world. So it starts here.

But to take it to the next level, make sure you go to your everyday AI.com. We're to be recapping today's show as well as giving you all. the other AI news and developments you need to know to be the smartest person in your company. So a lot happened this week. Absolutely bonkers. All right. So if you are brand new here, most Mondays, we do the AI News That Matters show. All right, but we do this everyday AI thing Monday through Friday.

So let's get you caught up. Like I said, even if you were paying attention, yeah, a lot happened this week. So let's get to. the AI news that matters. Well, first, Open AI announced their deal with the Pentagon after one of their chief rivals, Anthropic, had been publicly bickering with the Department of War and just the U.S. government at poll. So OpenAI CEO, Sam Altman announced the new agreement between OpenAI and announced the agreement with the Department of Defense that will allow the Pentagon to use Open AIs AI's AI models on its classified network.

So this development comes as OpenAI's rival Anthropic failed to reach a similar deal with the Pentagon, leading to escalating tensions in a directive from President Trump to phase out all federal use of Anthropics products within six months. So the Pentagon's push for quote-unquote all lawful purposes, access to AI models sparked some debate with Anthropic drawing a hard, line against uses such as mass domestic surveillance in fully autonomous weapons. All right. So more on the Anthropic versus, well, I don't know, the U.S. government angle here in a little bit,

but a little bit more about Open AIs deal and the response so far. So more than 60 Open AI employees and 300 Google employees signed an open letter supporting Anthropics stance urging their companies to prioritize. ethical boundaries in military AI use. So Altman emphasized that OpenAI's newly reached deal includes protections against domestic mass surveillance and requires human oversight for any use of force, including autonomous weapon systems, aligning with stated safety principles. Reuters also reported that Open AI's agreement includes a contract enforcement lever for OpenAI

that they can say that if a breach happens, it could terminate the contract kind of on their terms. So OpenAI said it will build technical safeguards and deploy engineers alongside Pentagon staff to ensure their AI models are used safely and as intended. Altman said OpenAI pushed for the same terms that was included in their deal to be offered to all AI companies, calling for de-escalation and reasonable agreements. that was included in their deal to be offered to all AI companies calling for de-escalation in reasonable agreements instead of legal battles. And obviously, this announcement came just before news broke of the U.S. and Israel government's beginning military action against Iran, adding to the urgency and scrutiny around AI's role in national security decisions.

So, yeah, this one was... not really expected, right? We're going to get to what happened a little bit more between Anthropic and the U.S. government. But kind of over the weekend, we got this announcement from Open AI that they had entered into an agreement with the federal government, which part of it was seemed like kind of out of nowhere, right? But it also makes sense because reportedly some big AI leaders

are going to be meeting with U.S. President Trump at the The White House this week, Anthropic, was notably not on that list. So maybe this makes sense. And obviously, a lot of people are like, oh, this deal came about very quickly, probably not, right? My assumption here, I don't have this on any, any source necessarily, right? But my assumption is these talks had been ongoing for a very long time. And kind of looking at some of the comments that Sam Altman made over the weekend.

He kind of did an AMA, you know, asked me anything. session in Twitter, kind of showing that there's been a lot of thought that has gone into this, and it wasn't just in response to Anthropic and the U.S. government not reaching a deal by the deadline that the U.S. government had imposed on Anthropic of Friday. So maybe it wasn't kind of the knee-jerk reaction competitive move that a lot of people think it was, although it might kind of seem like that on the surface. But if you know anything about the government and how it works in these huge contracts, you know that they generally take many, many weeks of work, sometimes more. So I'm sure that we'll get more reporting. And this is one new story that is

definitely not done. And oh, hey, I should shout out. Good morning, live stream audience. good to see adeno from italy uh dennis from north carolina joe from florida jose from santiago uh yeah thanks for thanks for joining us yeah if you listen to the uh the podcast maybe on spotify uh apple yeah this thing is also a live stream uh every single day 7 30 a m central standard time all right let's talk a little bit more about the other big ai news story of the week and that was well U.S. President Donald Trump has ordered every U.S. federal agency to immediately stop using technology from anthropic, giving agencies like the Defense Department a six-month window to phase out the company's products. This is huge news. All right.

So a little bit more on what the heck happened this past week. So the White House's decision follows a standoff between Anthropic and the Pentagon after the AI company refused to allow its technology to be used for fully autonomous weapons or mass domestic surveillance of Americans. So defense secretary Pete Hegesith responded by designating Anthropic a supply chain risk to national security, stating the Pentagon would transition to another provider within six months. So Anthropic, which signed a $200 million contract with the Pentagon in July, said it will challenge the supply chain risk designation in court, calling the move legally unsound and a dangerous precedent for American companies negotiating with the government. Anthropic also said the impasse was over two specific requested carve-outs to the lawful use terms, which were mass domestic surveillance of Americans and fully autonomous weapons. And this was not a blanket refusal of defense work from Anthropic side. So Anthropic responded to the Trump ban by stating it had not even received direct communication from the Pentagon or White House.

And they said they kind of just found out with the tweet or the post on Trump's truth social network. And they vowed to challenge this in court. So Trump accused Anthropic of putting American lives and national security at risk by insisting on its own terms of service and vowed never to do business with the company again. So the dispute highlights growing tensions between big tech companies and the U.S. government over the ethical use of AI, particularly for military and surveillance application. Also, there's been some noteworthy reaction to open AI's deal and inpropics refusal. Because over the weekend, there was a ton of, you know, not just talk.

I'll call it much, much more than that. There's a lot of online kind of backlash against open AI for entering into this deal with the U.S. government and a lot of support for Anthropic refusing to do so, so much so that you have, you know, kind of noteworthy names in the AI field. You had celebrities, right, posting about this, but it did actually, not that this is important, It did catapults in Fabix, Claude app to the top app in the U.S. on the Apple App Store, which hasn't happened before. What also hasn't happened before is, I believe, is a U.S. company getting the supply chain risk to national security designation. This is an enormous detail.

Again, I'm sure we're going to be talking about this a lot in the coming weeks and months and what happens because this definitely will be challenged in court. But essentially what this could mean if it is fully enforced that Anthropic, well, could in theory lose a lot of its current business deals because if current companies who do work with the federal government are still using Anthropic, well, that means that they would no longer have those deals or that work with the federal government. That's what this means if this is actually enforced, which could impact 100,000, of companies, right, that both have contracts with the federal government, but obviously still use Anthropics models.

So this is going to be, I think, a straight up fight, right? We'll see what actually happens from it being enforceable because, you know, like I said, Anthropics said that they didn't even really get official designation of this. They just got, you know, word like everyone else via social media posts. So we'll see. I do expect this to go on for quite a while. All right. Well, now getting away from super serious stuff that affects, I don't know,

the global piece and, you know, maybe AI making decisions on the battlefield. Well, let's get into things that are maybe less important, but still noteworthy if you follow and work in AI. That's because Google has unveiled nano banana to a new. AI image generation model that delivers high reasoning, accurate text rendering, and creative control at half of the cost of its previous pro-tier model. So the new nanobanana 2 is built on the Gemini 3.1 Flash backbone, offering images in the API at 50% or half of the cost of the nanobanana pro model with better results.

So Google's new Nanobanana 2 model took home the top benchmarks for AI generation on Arena, formerly LM Arena, which had previously all been held by OpenAI. Yeah, a lot of people don't know that, right? Obviously, Nanobanana went super viral, Nano Banana Pro. And then with Nanobanata 2, well, actually, Google was technically behind on some of these head-to-head blind test benchmarks. and the general internet, at least slightly preferred open AIs image generation models, well, not anymore after Nanobanana 2. So right now, this is available by default, which is pretty big. In Google apps like Gemini search ads and even the new flow product for most users worldwide,

as this has rolled out to 141 new countries already and is available in eight languages. So if you are a Google Gemini subscription, whether on the pro or ultra plan, you can access Nanobanana Pro for specialized tasks, just straight inside of Google Gemini. So Nanobanana 2 supports full aspect ratio control with resolutions from 512 pixels all the way up to 4K and introduces two quality settings to balance speed and image fidelity. So Nanobanana 2, some of the new kind of features or big upgrades, it's significant improves adherence to complex prompts, capturing nuanced requests more reliably than prior models.

So the model also now enables accurate, legible text generation and in image translation, which is pretty cool, addressing a longstanding weakness in AI generated visuals and can maintain subject consistency across up to five characters and 14 reference objects. And one of the coolest things of all of this, which is like one of those little footnotes, but I think is really powerful, well, there's a new image search tool, which allows nanobanata 2 to retrieve and ground images as context for generation from search, expanding its utility for complex enterprise workflows like storyboarding and product photography. So, I don't know, live stream audience, I'm always curious.

Are you guys interested in podcast audience? You know, feel free to leave me a comment. I go through and check them all on Spotify. Are you guys interested in learning more about nano banana too? I'm always on the fence, right? Because when the original, you know, AI image generation boom happened, I don't think that the actual models and capabilities were useful enough for the everyday business user, right?

Obviously, if you work in marketing, content creation, right, the original nanobanana and, you know, Chad Chubit image were great. But I don't think it was really for the, you know, non-creative. But I think that's really changed, right? In the past couple of months, in terms of what you can accomplish with these models. And I think nanobanata too obviously takes that to the next level, right? Even with things that I just said, like better text rendering, being able to ground images in Google search. people are thinking, oh, it's just, you know, some high fidelity, cool looking image. No, right?

This is obviously something that can power infographics, individual learning, right? It can really, I think, change how your company communicates with consumers whose attention spans are obviously getting shorter and shorter and who demand, you know, well, let's be honest, more personalized visuals. And I think now we finally can get AI images that don't look AI, where, you know, six to nine months ago, even though the models were great. they still had a very AI stink to them, right? Not in a bad way, but let's be honest. All right.

Angie at least says yes to more nanobanana. So does Jonathan. All right. We'll see. Maybe we'll do something in the coming weeks, just a non-technical or non-creative guide to nanobanana pro. They're sorry, you know, banana two. 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-plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premier, Lightroom Express, and more to help

bring your ideas to life. You can also get started with creative skills, a growing library of 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. dot Adobe.com. The updates come so fast. It's even hard to keep track of the names. All right. Well, here's something unlike AI images that may or may not affect every single business user. This one definitely will.

So Open AI has announced a new multi-year partnership with four major consulting firms, Accenture, Boston Consulting Group, Cap Gemini, and McKinsey and Company to help deploy its enterprise AI platform Frontier. So the move marks a significant push by OpenAI to scale up its business with enterprise clients, which already accounts for up to 40% of its revenue. So Frontier, which was unveiled earlier this month, is designed to integrate and manage AI agents across company systems, making it easier for businesses to automate tasks and harness AI more efficiently. So these consulting firms will work directly with Open AIs engineers and form dedicated, certified teams to implement the technology in real world business environments. So Open AIs chief revenue officer, Denise Dresser, says the partnerships leverage consulting firms deep relationships and expertise with large clients helping to meet the high demand for AI solutions.

So the consulting partners will receive access to Open AIs product roadmap. technical resources and research teams ensuring they stay at the forefront of AI developments. So this alliance comes as Open AI competes with rivals like Google and Anthropic for Enterprise Market Share, intensifying the race to lead in AI adoption. So not to say, I've been screaming about this for a very long time, but I've been screaming about this for a very long time. And it's not surprising at this point, at least, maybe it's. It was surprising if we had this news headline 18 months ago.

But I've already said this. The consulting industry is going to get wrecked, right? Completely. And it might look more like this, right, with big consulting firms, the global juggernauts, right? Like Accenture, Boston, you know, BCG, Cap Gemini, McKinsey, et cetera, entering into these huge agreements. What you should be keeping an eye on, right? Not just if you work at one of these companies, maybe you engage them for work, right? Are we finally going to get away from the billable hour from consulting groups?

Are we going to be moving to an output, you know, getting charged for outcomes? I think probably, right? So take a look at what these four companies do in terms of how they communicate about their AI offerings and services because the reality is, and I've been saying this, today's AI technology, I don't think most people, unless you listen to the show every single day, unless you're using the technology every single day, I don't think most people realize that, well, today's agentic AI systems are much better. Yes, they are much better than the average skilled human worker

at even producing outputs, right? And in the consulting space, let's be honest. A lot of what's done there is research, putting together spreadsheets, gathering insights, putting together PowerPoint decks, right? These are all things that off the shelf models can do by default with a single prompt, right? So you can make the argument and there are, you know, kind of blind benchmarks that prove as much that basic models that all of us have access to are better than human experts. So it is going to be telling, I think, what comes out of this new partnership. the frontier partnership with at least these four big consulting firms.

And make sure you go back and listen to our 2026 AI prediction and roadmap series. Because I talked about this specifically. I haven't talked about it at the end of 2024 in my 2025 AI predictions. But I do think that this is going to be the next big straw that shakes up knowledge work in general. Because the reality is whether you work at a consulting company or not, I don't think that at at this point next year, people will still, right, even in industries like legal, accounting, finance, et cetera, I don't think that companies are still going to be able to charge that same price, that same high hourly rate that has come to define the last generation of knowledge

work. It's going away. So pay attention to what happens with those four companies. All right. Our next piece of AI news, which technically happened like six and a half days ago, But it feels like six and a half weeks ago, but we have to mention it. So this is big. So a major new report from Anthropic has far global concern over the security of American AI innovation, as Anthropic claims three top Chinese AI labs extracted advanced capabilities from its clod model illegally through distillation. So Anthropic publicly accused Chinese companies deep seek moonshyke, Moonshot in Minimax, three of the leading Chinese AI labs, of using 16 million prompts in more than 24,000 fake accounts to extract and copy Claude's most advanced features.

So the labs reportedly use a technique called distillation, which means they're training their models on outputs from Anthropics models. effectively cloning Claude's capabilities for the Chinese companies at a fraction of the cost and time. So Anthropic said that this activity was detected through IP addresses and metadata, with Anthropic stating that one proxy network managed over 20,000 fraudulent accounts simultaneously to avoid detection. Anthropics not alone, as we talked about in our episode last week when we covered this, as Open AI and Google in the same like two week period, both reported similar attacks with OpenAI sending a memo to Congress and Google identifying over 100,000 distillation props targeting their Gemini model. So it's important to keep an eye on this story because Anthropics market share on open router has dropped sharply from 40% a year ago to, well, it depends on. the day you look at it, but more like 9 to 13% today, a loss that could represent hundreds of

millions or even billions of dollars in annual revenue. And as Anthropic prepares for reported IPO in 2026, this is going to be huge because even though we've seen recent reports that Anthropics monthly revenue and annualized revenue is still growing month over month. It's growing at a much slower rate. So I think that we're starting to see some of the leading factors of that, such as, right, their open router share has gone down significantly. So open router, you don't know, well, they are the leading third party provider. So for the millions of companies that are building AI into their products and services, well, a lot of them use open router. It allows you to more easily use multiple different services, swap, you know, infropic in for Gemini, Gemini

for Anthropic, vice versa, right? So they have great data on where the industry is heading. And Anthropic, for the most part, has been the leader on the API side, the developer side. And we've seen now these new Chinese companies with their models have become leaders in the space. But Anthropic is saying, well, they're only leaders because they copied our models. They distilled our models. So distillation is actually a very common practice when it comes to the company itself. Right. So we've heard from all the major AI labs saying they use their big models.

And then they use those big models to distill smaller models that then they use and they produce those, right? But it is not normal and technically, well, legal for other companies to distill from American companies. So right now there's really no international or even domestic laws preventing this kind of model distillation. and enforcement is especially challenging when foreign actors are involved. So the U.S. government has attempted to slow China's AI progress through chip export controls, but distillation offers a fast and cheap workaround for Chinese labs to catch up or surpass American models. And y'all, I talked about this last week, so go listen to Episode 720 if you want the deeper dive in this. But this impacts everyone.

This does not impact you just if you work at one of these. companies or if you're using these models, you know, to build something at your company. One of the biggest reasons is this is going to have, if this continues at the pace, right, like I said, it had kind of been a well-known fact, but it wasn't really talked about until early 2026. And in February, right, we got public statements and confirmation from the big three players in Google Open AI and Anthropic. And if this continues, what this essentially does, well, you have the biggest AI labs then doing all of this work.

And then their Chinese counterparts essentially undercutting them and being able to, you know, provide models that have 99% of the capabilities for 1% of the cost in 1% of the time. So when we talk about how AI is used across military applications, which is just like what we talked about earlier. So not only does this take away a competitive advantage for American AI companies, but this also provides other countries who may not exactly be on friendly terms with the U.S. Access to technology that will likely be used to power their military decision making. Right. And that's where I think a lot of the focus has fallen over the past year or so. And I've been saying this for a very long time. Artificial intelligence and having access to the most capable models is more important.

maybe this doesn't seem as crazy when I first said it like two or three years ago, right? But access to the best AI models is more important than a company's military, their access to natural resources like gold and oil, all those things. Having the best AI in the world is much more important because it does lead to artificial general intelligence. It leads to artificial superintelligence. And whatever companies can get there first have essentially an unfair advantage when it comes to being the future global superpower. All right.

So now let's get to more things on the agenic side. All right. So we have a lot of stories here that all happened at once, which was surprising. So let's first talk about Microsoft. So they announced co-pilot tasks. So this is a new AI system designed to handle time-consuming tasks for users in the background. So Microsoft says that copilot tasks offload work from your device to Microsoft's own cloud-based computer and browser, freeing up your machine for other activities. So,

yeah, essentially scheduling AI agents to go do your work for you and not having to clog up your own computer's resources. Sounds awesome, right? But unfortunately, hardly no one has. access. So co-pilot tasks is currently in a limited research preview with access restricted to a small group of testers for now. And you can, you know, go sign up to be on the wait list like I did. But I'm guessing that this is going to be a very, very slow rollout. But looks pretty amazing. So users can assign co-pilot tasks to handle jobs like scheduling appointments, generating study plans, organizing subscriptions, and even canceling unused services. So the AI can turn emails, attachments, and images from your inbox into slide decks, and it can

surface urgent emails and draft replies automatically. So co-pilot tasks is capable of planning events, such as birthday parties from booking venues to sending invitations, and can monitor apartment listings and set up tours on a recurring schedule. So Microsoft says users can describe what they need in plain language and assigned task can be completed once on a schedule or as recurring tasks. The system will always first ask for permission before taking significant actions, such as making payments or sending messages to ensure user control and security.

So this was not as like awesome and amazing as that sounds. this was barely the third biggest piece of news or AI developments over the past week just when it came to agentic AI features. All right, because our next piece of AI news probably way bigger and the availability is, well, more available. So Anthropic has introduced a scheduled task feature for Claude Co-work in the desktop app that in enables users to automate complex multi-step tasks using AI. So this works a little bit differently than Microsoft tasks. So some overlap in terms of features and some unique features. So here's how it works.

So Claude Co-Work now allows users to schedule recurring tasks, such as generating daily briefings based on Slack messages and emails, conducting regular competitor analysis, or even just organizing files on your computer. So the service runs in a dedicated virtual machine separate from the main operating system, enhancing security and preventing unwanted access to sensitive files or system functions. Users must give explicit permission before Claude Co-work can take any major actions, like deleting files and can control which third-party app integrations the AI can access. So to use scheduled work tasks, users input a task prompt,

and they can specify the cadence and instructions, which Claude then executes autonomously. Here's the downside as long as the desktop app and computer are awake. So, yeah, Microsoft's new co-pilot tasks can take certain tasks to the cloud, right, but still work with your local data that you give it access to. Claude Co-Works, scheduled tasks a little bit different. You have to actually have that computer and that, instance of Claude Co-Work, the desktop app running. Because if the computer or the app is inactive, those scheduled tasks are skipped and will

resume once the device is active again. So task management, including pausing, deleting or running schedules on demand is handled through the app's left sidebar. So who has access to this right now? Well, a lot more people than Microsoft tasks. So the research preview of Claude Co-Work is now available to all. all paid subscribers, including Pro Max team and enterprise plan. So let me know, have you guys been using Claude CoWorks, scheduled tasks I have?

Right. I don't think my poor Mac Studio has slept since OpenAI's Codex app came out like three weeks ago. I think I've restarted my computer once, but it's running around the clock, right? I'm not just the, you know, Claude Co-work with the scheduled task, but obviously I've been super, super loving codex and the new GPT-53 Codex max model. Yeah, if you're not just vibe working around the clock, you're definitely falling behind. All right. In our last piece of AI news, well, at least our big stories, more on the agentic side. I told you this week was all about AI drama in the government and new agentic offerings,

because they all just came at the same time seemingly out of nowhere. This one from Perplexity, I haven't gotten a chance to use it yet, probably this week, but it could be maybe the most impressive. We'll see. So Perplexity has unveiled Perplexity Computer, a new cloud-based agentic AI tool designed to function as a digital co-worker. So this new tool stands out by performing complex, multi-step tasks using several specialized AI agents

that collaborate to deliver finished products. So the cool thing about a perplexity computer is it can use, I believe, 19 different AI models from all of the big players. Also, it leverages multiple AI models at once, including Claude Opus for logical reasoning, Gemini for research, you know, other models like GPT52 for more fact-based tasks, and other models for images, video, and rapid subtests. So unlike similar agentic AI solutions that run directly on users' devices, like

Scheduled Tas and Cloud Co-work, for Plexity Computer operates entirely in the cloud, reducing security risks to local PCs and files. So yeah, this is being built more as a more secure open claw competitor, right? That's been the narrative so far, right? You can see what these big AI companies release and how they market it and how they position it, right? But it's really kind of AI users reactions to how they're actually using it in their day-to-day tech stack. And it seems like at least original initially that perplexity computer could have legs here.

Because a lot of people are saying in terms of capabilities and what it can do, comparing it to open AIs open claw. Yeah, that's still weird to say after Open AIs acquisition in February of the very popular open source, autonomous ancient technology, open claw. So a little bit more about computer. The system is capable of creating dashboard. building apps in generating presentations by delegating different parts of the work to its subagents. The downside right now, well, it does require that very expensive perplexity max subscription,

which is $200 a month. So is this something that could take, you know, over the narrative in terms of, you know, people's favorite, you know, AI agent, maybe? But I think when, you know, users are looking at their subscriptions pile up, right? You know, the $200 plan from Google Gemini, the $200 plan from Open AI, the $200 plan from, you know, Claude, right? I don't know. Is there really room for another, you know, $200 a month plan from perplexity? Again, in terms of the capabilities, once you use all these things, it's like, oh, yeah, of course, right?

But when it comes to businesses wanting to, you know, implement this at scale or, you know, entrepreneurs, solopreneurs, you know, choices can be limited, right, as the different subscriptions start piling up. But regardless, I think perplexity computer could be their most important product. All right. More than two years ago, I predicted that perplexity is going to have to pit it hard or they would get absolutely squashed. And, well, they did, right?

both with their comet browser, which I think is really good, their model council, which kind of uses this different approach to have multiple models, working on a problem at the same time. And now we got computer from perplexity. So they are no longer just an answers engine because all of the, you know, Claude, Gemini, Chad, JavVT do that better, I think by default. So perplexity here, I think with a pretty good offering. So something we'll be keeping an eye on.

All right. And yeah, if you need to get caught up, FYI, this was a very timely start here series episode. You're right. So we've been doing the start here series throughout, you know, January, February and March, one to two of these episodes a week. Because if you don't know where to start in AI, I think this is a great place with that series. So actually, volume 10 was great. Make sure to go check it out.

That's episode 723. You can also just go to start here series.com. We included all of these three updates from, Microsoft, Anthropic and Perplexity, and just as we took a look from how AI has gone from chatbots to autonomous worker.

So if you haven't checked out that episode, make sure you go to episode 723 or just go to start here series.com. And yeah, we're probably going to be tackling one of these on our Wednesday show. So on Wednesdays, we do this, you know,

putting AI to work at Wednesdays. I'll probably rerun the poll in today's newsletter as well to just see what do you guys want to hear more about on those three big ones because it was actually a pretty close kind of race when we asked you guys on Friday. All right. Those are the big AI news stories. Now let's quickly do a wrap with what's new and what's next. So if I'm being honest, just our bullet point roundup is more consequential and bigger AI news than we would have gotten like six months ago. But here's everything else.

some big AI news stories that didn't make our, you know, top eight cut, some leaks, some rumors. Let's get into it. Bullet point style. Here's what's new and what's next. So the U.S. military reportedly used Anthropics models in its strike against Iran, despite the recently enacted government ban. So that may not have lasted for long. XAI struck a deal with the U.S. government for classified use.

Block, right? run by former Twitter CEO, Jack Dorsey, cut 40% of its staff or 4,000 jobs, citing AI productivity. Google Labs upgraded Opel with autonomous agents, tool calling, memory, and routing. My gosh, Opel's so good, so easy to use. Cursor released cloud agents, really cool. It checks your work and sends you a video. Meta committed $60 billion to a multi-year AI chip deal with AMD, and they also secured potentially a 10% equity stake in the company.

NVIDIA announced very strong quarterly earnings with $68 billion in revenue, but shares fell on AI spending concerns. Meta is close to releasing their AI video editor, Bibes, to compete with SORA, according to leaks. OpenAI hired Meta AI chief, rooming paying, who was formerly on a $200 million plus package with meta after meta poached him from Apple. So now he's been re poached after spending less than a year with meta. So gosh, this developments are crazy.

It's like, you know, NBA players. All right. President Trump is set to meet Tuesday with AI and tech leaders to address data center power costs and infrastructure demands. All the big AI companies in the world except Anthropic was. notably absent from the list, right? Obviously, IBM shares fell more than 13% after Anthropic claimed AI automates cobalt modernization. Yeah, Anthropic came out with some impressive enterprise plugins this past week.

Open AI expanded GPT53 codex access via the API. So now expect some benchmarks to come in via that and all the other, you know, vibe coding platforms just got a whole lot better now that they can use GPD 53 codex. Infraopic, like I said, added enterprise plugin marketplace that gives admin controls, new connectors, and Excel PowerPoint orchestration. Google acquired producer AI to expand AI powered music composition using Leria 3. Anthropic launched Claude Code Remote Control, enabling mobile management of desktop coding sessions. I've been enjoying that.

I don't have to, you know, log on to, you know, my Google Chrome remote desktop on my phone. I can just launch the Claude code remote control. Claude also rolled out a new plugin to connect Slack and cloud code. Microsoft introduced a sovereign cloud tool to enable offline secure AI infrastructure operations. OpenAI introduced GPT real-time 1.5, improving voice command accuracy by 10%. Notion launch custom agents and Google added some pretty impressive templates for VO. My gosh, that was a ton. So if you spend hours every single day trying to keep up with what's happening in AI and you feel like you're barely treading water, stop doing that.

Number one, read our newsletter every day. It takes like seven minutes and then join us on Mondays as we break down the AI news that matters. That's it. That's a wrap. A very busy and impactful week of AI. These are obviously some storylines that are going to continue to develop this week. So make sure you do go to our website at your everyday AI.com. Sign up for the free daily newsletter. Thank you for tuning in. We'll see you back tomorrow and every day for more everyday AI. Thanks y'all. Meet Firefly AI assistant. Now live in Adobe Firefly, the Allman One creative AI studio. Just describe what you want to create in your own words and the assistant handles the rest. Orchestrating multi-s

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. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. 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.