Anthropic beats OpenAI? + 5 AI Stories You Missed

0:00Anthropic is beating OpenAI in revenue, but that also means that they're taking money away from users like you and me. We got a solution for you. Codex now can control your computer and one user had it fight Amazon and win him a refund. Grok is now enabled in Hermes agent and it can superpower your social media. We'll show you how. Google is coming out with a new laptop that's an AI first laptop. Oof. One stupid thing about it, one useful thing. We'll talk about that, too. And finally, why are people putting their fingers in Mac books when there's a terminal command for that? All that and so much more coming up in this

0:34week's news. Let's get into it. I think the best news story this week is that OpenAI's Codex can control my computer. In fact, here, this is from their demo. I'm going to show how they're using Codex and telling it to go and create a reminder in the reminders app, because whether it's from your desktop app or

0:51the phone, you can now control apps. What do you think of this? So, I think it's a huge update, right? Computer use isn't a new thing, but they built it right into the Codex app. And with Codex becoming a lot more popular, I mean, I feel like the vibe and kind of the sentiment of the community has shifted away from Claude code and toward Codex. There There's going to be so many cool use cases that come out of this feature. And this is, you know, what we're about to see is one of probably a thousand plus use cases that you could

1:17come up with. I think so, too. My top use case is I can't get YouTube videos easily without using a browser and computer. I can't get X data without using a browser, and so I use a browser all the time. My big disappointment with all this is that I need to keep my computer going. I don't want to have a computer going. I want

1:40the cloud again. That seems to be a big pain point that people are having. I mean, before we hit record, we were looking at different like crazy ways that people were using to keep their computer running while the while their MacBook was closed, for example. If you've got like a Mac mini like I do, I mean, obviously you can just set it to never sleep. But, for folks that are vibe coding on MacBooks, there are workarounds, but they're a little janky. Like, you think that would be a built-in feature or something that

2:04you could easily set up. about something like this, which will as you showed. He's He's using this little dongle, connecting it to his computer, and keeping the Mac open and on even when the lid is closed. By the way, Windows people are laughing at us. I will say as ridiculous that as that is, I'd rather carry my laptop that way than put a finger inside it because I've got it doing something. And Just the other day, I I was going out for a drink with my wife, and on the way over, I needed to finish something. And so, I um And so, I I had to do something like

2:36that. Having said that, I think this is unnecessary. There are multiple ways to do this. I think that there's something called Amphetamine. Well, the fact that it's 2026, like we literally have LLMs, and we don't have a way to keep our MacBooks It It's just I I hate that we even need it. Here, this is Amphetamine. It's the MacBook app that's free that will do it for people. And here's um And on Reddit, you can see that people are showing that you can clearly use a terminal command to keep it on when the lid is closed. I still think that the big issue is I don't want to

3:07have to keep my computer on for all these tasks. But, if we need to, it's great to know that now with Codex, we can do it, and it works really beautifully. I want to show a great use case here. This is um Jason Lu. He is on the Codex team. I love what he did. He used at computer. This is how in Codex you get to fire up the command to use the computer. And he said, "I want you to wait on hold for me with Amazon for 25 minutes because a package was stolen, and I want you to talk to a human being." Here's the

3:37prompt that he gave it. And he essentially said, "Once the human being comes on, I want you to start negotiating with them." And by the time he got out of the shower, the refund was done. This is another great use of Codex. Beyond Codex. And this was in a a a support agent chat on the website,

3:54right? This wasn't like a voice chat? Exactly, yeah. Right. Yeah, I mean that's again one of probably a thousand use cases you could come up with. I mean there's so much nonsense that you have to do inside the browser. I mean even think 30 minutes ago before we jumped on here and hit record, I had to file my annual report for my LLC. And that's something that's very straightforward. Like there's not much to that. I I so easily could have fired up Codex, had it take over my computer, submit that and be done. You know, it only took me 7 minutes, but that's 7 minutes that I didn't spend on

4:27a sales call or, you know, writing a tweet, stuff that could actually make me money. Here's what I'm going to use it for. There is a site that you can use to reduce junk mail, you know, junk US paper postal mail. And it's so annoying to go through. I'm just going to tell Codex to go through

4:41and do it for me. All right. Other uses of of Codex here from Jason, I love. He's basically using it as a chief of staff. He's saying every 30 minutes check my Slack and Gmail for unanswered messages that need my attention. Help me prioritize what matters most. If someone asks a question, research the answers deeply as you can and draft it. He's finding that that often the draft reply from him is just as good

5:03as the one that he's sending himself. It allows him to monitor feedback. Obviously, I talked about getting a refund. Um he's just using it for a lot of personal use, not just for coding. Codex is really done well and it's uh getting a lot more traction with us after it was crushed by Claude code. By the way, I'm sponsored by Zapier which I use to connect about 10,000 apps to any AI agent that I use. I mean, when I'm in Codex, I've got it. When I'm in Claude, etc. It's what allows me to bring all the apps that I want with the restrictions and protections that I want

5:36into every agent and tool that I love. Go check them out. They're MCB is fantastic at zapier.com/mcb. All right. Speaking of Open AI, here's the next thing that I think we should talk about. The next news story is that they have released this new finance feature. Now, in ChatGPT, if you have one of the pro accounts, I think it's $100 or more, you can now start to import data from your bank accounts. I love that you can connect directly to them here. This is Charles Schwab, it's American Express, etc. You import data in and you can start asking questions. I freaking love that it's doing this. What

6:10do you think? Yeah, so it's so funny that they're doing this. I actually saw this coming from a mile away because Perplexity has had this capability for probably the last three, maybe even four months. Now, Yeah, yeah. And so, I'm not sure if ChatGPT's connection goes through Plaid or if these are all like native bank and financial services platform connections through ChatGPT, but in Perplexity, you know, you connect to your bank, your credit card, wherever through Plaid, and then you can pull in all your financial information and basically do exactly

6:42what ChatGPT is saying it can do here. Now, obviously, what ChatGPT did is they saw Perplexity do that three or four months ago. They saw it really elevate their platform. Like, a lot of people joined Perplexity to have that ability, and ChatGPT has a distribution to literally copy and paste, plug that into their existing user base, and I'm sure people are now leaving Perplexity to come back to ChatGPT just for that feature. So, they're just going to sit back and watch what works for other labs and then copy it because they have the

7:11most distribution. It does look like they are using Plaid in order to do this, and yeah, I guess you're right that all they have to do is see what everyone else is doing, and because they have such a big user base, they can copy it. That hasn't worked for Google with Gemini, but we don't But, Gemini is becoming a mess. All Google products are becoming a mess. With ChatGPT and Claude, there's like a

7:31streamline effect here. Um all right. This is one way you could do finances. Um in related news, Claude just announced this, and you've actually pointed out that this is not really a new announcement even though this came out like what was it 6 days ago now so it's almost a week old. They've said that they can that they will use their um Quick QuickBooks connection there it is the Intuit QuickBooks connection to let you analyze whether you've got enough money in the bank for payroll to help

8:02you understand your company's finances. Let me scroll through this this video and and show some of the charts. It goes into spreadsheets. I still don't like this connection. This one's pretty terrible and I and I have a few reasons why I think it's bad. Well, it's so funny to me because this is so clearly a marketing play on on the part of Anthropic like Claude has had a QuickBooks native connector for I don't know how many months at this point. So all this is is Anthropic probably saying like oh we don't really have a big

8:33update to ship this week or this month. So we're just going to take a feature that we've already had package it into a a neat little announcement and position it as a you know Claude for small business suite of tools that literally every one of them already existed. So again is it helpful to small business owners for sure. I mean who doesn't want to get visibility into their P&L and their payroll and be able to analyze their finances in their existing tools with natural language but this has already existed. This is just a

9:03marketing spin in my mind. I I also don't like it because this connection is not very good. QuickBooks is not looking to be generous with their data. They have very big limits on how much data you can get into here. They are it's a slower connection. I actually spent the weekend trying to figure out my my finances. I really love being able to ask Claude or ChatGPT for data. I wrote this up about how I did it. I found that my best connection my best approach was to use SimpleFin for a buck 50 a month. I can get all the finance accounts that I have put through

9:37SimpleFin, then SimpleFin sends it over to Superbase, just a basic database that I use for free, and then I use Claude to chat with it and ask questions like, "Tell me where I'm spending a lot of money. Tell me about my recurring expenses here." Put together my recurring expenses for the family. "Tell me where we went wrong last month. Why is it that suddenly our expenses are high?" It created this beautiful chart, which by the way, I don't know that I want to call this beautiful anymore. Claude has their own

10:03look, and I'm I'm getting tired of it. I was about to say I can very easily tell this is a Claude design. Like Claude design generated design, you know? I hate it. It's so annoying. It's tiring. We've seen it everywhere, but okay, it's on me to pick a different one. But if I was writing about it, I wanted people to see, yes, it is a

10:19Claude thing. It's not me faking it. Having said that, this is great progress. I want them to have more information about our finances, um and I want to I want more help with my finances, and I find that AI is doing a great job with that. Right. In other news, it looks to me like what we are seeing is that Anthropic is now basically neck and neck with OpenAI for revenue. In fact, they've actually inched a little forward. This is from The Information. Anthropic can claim about half the amount as its revenue recently surpassed OpenAI's by half the amount that means like basically they're saying that um the majority of revenue

11:00in AI is going to Anthropic and OpenAI. I thought that was uh interesting and kind of obvious, but the part that stood out for me is we're seeing that Anthropic's revenue now is passing OpenAI's. And in a moment we'll see what what's happening in small businesses, too. But this is this is interesting. It's super interesting, and I feel like you know, if you if we were to rewind, let's say about a year, everybody was pretty much Well, maybe even more than a year. Year, year and a half ago, everybody was on the OpenAI ChatGPT

11:30train. Just because, I mean, they were the 800-lb elephant in the room, right? They're pretty much the only AI provider that anybody knew about. At the time, Claude was ranked like 160th in the in the Apple App Store, something like that. And it really wasn't until I want to say the the OpenAI deal with the US government went south, or sorry, the deal with the US government and Anthropic went south, right? When Dario Amodei uh walked from that deal because he didn't want to give the government what he thought were, you know, was oversight that that wasn't warranted with using Claude's technology, and OpenAI swooped in and took that deal. I think it was

12:10that moment in time that the public sentiment shifted more away from OpenAI back towards Claude. And, you know, I remember it specifically, Katy Perry tweeted a screenshot of her downloading the Claude app, and it got like 10 million views. And I remember thinking, I'm like, okay, if Katy Perry is talking about using Claude, this thing has gone crazy mainstream. It went from nobody in the world had heard of Claude except the nerdiest of nerds like me and you, to, you know, 30 days later, that's what everybody's using. That's all you're seeing on social media. And the vibe was heavily skewed towards Claude code, but now it's kind of shifting back towards

12:48OpenAI and Codex. That's something I've noticed. I think that geeks like you and me and whoever's listening to us right now, we're the ones who really are at the forefront and determine what's happening in the future. It just took them a little while to catch on in mainstream, and what did it is, like you said, that one news story. But they were killing with Claude code. The AI that can do more than answer questions, can actually do stuff for you. They built it up, they got us excited about it. We then were I I mean, not to pat ourselves on the back, but we are we are the tastemakers when it comes

13:17to technology, all of us nerds, and that's that's what happened. Look at this. This is from Ramp. They basically look at all of us Ramp customers to see where we're spending money and what we're what they're seeing is for us Anthropic has already beaten OpenAI but the reason I wanted to bring this up is not to just show like another group of people within this big group who are using Anthropic more. What's interesting here is they're also noticing that a lot of companies are now starting to use the free models

13:51more. And that actually is an eye-opener and I start I need to think about how I could start using more of the free models and how I could start creating automations that will shift my my stupid tasks to the cheap models and the the complicated ones to the expensive one. Up until now what I've been doing is just saying everything should go big and in fact

14:10this is what they've been saying. Anthropic's incentives are misaligned with business customers. They want us to purchase the most tokens. They're incentivized to drive us to more expensive models even when cheaper models are sufficient. We've seen Uber CTO say that the company already blew through their 2026 AI budget. Anthropic's going to lose market share if cost cutting drive firm if cost cutting drives firms to identify and route towards cheaper models. For what it's worth, this is also true for

14:38OpenAI. Um but they're already starting to see people use the cheaper models here and find ways to to incorporate them and I think that's an interesting eye-opener and we should all be doing more of that. I agree and so what's what's interesting to me right as you think about the everyday the average business user or the average AI user right there they're

15:00probably in one ecosystem or the other. They're using Claude chat or Claude co-work or they're using chat GPT or Codex, right? Now obviously when you're in one of those two ecosystems you're locked down to those models. You can't go and use an open-source model or you can't use a chat GPT model within Claude code for example. At least I don't think you can. Now, where it gets interesting, right, is for the folks like you and me who are using these open-source agent harnesses like a Hermes or like an open claw

15:29Mhm. and using tools like open router, right? I mean, I know when I first when I first started using open claw, I had no idea what I was doing. And at the time, you could run your Claude subscription through open claw, so I was just running Opus on everything. And once they took that away and I was spending 30, 50, 100 bucks a day on API costs running Claude models through my open claw, I was forced to use something like an open router and actually get get, you know, tactical with deciding when am I going

16:01to use which model and for which tasks. So, you know, for the folks that are using the open-source agent harnesses, something like an open open router to, you know, route lightweight tasks to like a mini max and then heavier tasks to like an Opus or a GPT-5-5. Last month, some of the fastest growing vendors on Ramp's platform were AI inference platforms that give companies access to cheap open-source models. Um All right, that is that is where we're going with it. Um let's take a look at this next news, which is that Super Grok is welcoming Hermes. Unlike Claude and Anthropic in general, the the Grok team is saying bring on the

16:41agents. What do you think of this? You're a big Hermes user. I think Elon knows exactly what he's doing. I mean, he if anything firsthand, he sees all the Hermes hype on the timeline. He knows that the sentiment is away, you know, shifting away from open claw towards Hermes because it simply works. It's not

17:00as finicky. It doesn't break as often. So, why would he not make Grok more easily compatible with something like a Hermes and try to gain market share for that small niche of the ecosystem that's that's going deep into Hermes, people like me and you. And uh you know, when it comes to this now being available, I think it was Boring Marketer had a really good tweet today. It was actually like a screen share. He He did a a recording of his iPhone talking about how he incorporated Super Grok with his Hermes agent, and he was just talking to his Hermes agent and letting his Hermes agent go and do research on X for

17:37trending topics in his niche, right? So, he was asking his Hermes agent, "Hey, go find trending topics of people talking about AI for marketing." And, you know, traditionally, it's been notoriously difficult to get your AI agents to pull information from X because X has a lot of anti-scraping and anti-bot measures. But, now that this is built in, I mean, it pretty much opens the opens the opportunity for these open-source agents to go out and and be on X, or at least

18:05scrape information from X a lot easier. Here it is. I just set up Super Grok with Hermes. This might be the biggest cheat code for social growth right now. He's saying, "What are the what use cases are discussed on X around autonomous agents, especially around I missed it." And then it's showing it. This is the example also that was on the screen before this that I was showing, which is it is really good for getting information from X. And, um, that has been incredibly difficult. All I do is find workarounds for how to get

18:32access to X information. Um, and so I I get where that's coming from and why that's appealing. All right. We're going to close it out with one final news story. I don't know exactly what to make of this, but I think it's interesting. This is the Google team. They're coming out with Google Book. I really like Chrome Unboxed because this guy is a passionate

18:52user of uh Chromebooks. And I thought Chromebooks were the future, but they're so freaking slow. They're now associated with stupid, slow, and crippled computers that kids get from school. And so, they're coming out with a brand new thing called a Google Book, not a Chromebook, but a Google Book. It is putting AI and Gemini of course specifically right at the forefront and it's got this thing that I thought was interesting here, which is you shake your mouse when you're looking at the screen and then it is aware of what you're looking at and it suggests some kind of action. The example that they had given was you've got a calendar

19:28invitation, you shake your screen and it says, "Okay, do you want me to put this on your calendar?" Or somebody brings up a date you want me to put on your calendar. Really [snorts] interesting. You can also then take these things that it sees on the screen and turn them into widgets, which I am not a huge user of Chrome Chromebooks. I didn't realize that they didn't have widgets. Um but I take his word for it. He is he's all over this. I didn't realize that you couldn't do a lot of things with with Chromebooks. This is adding it. I also will say that the other nice thing is

19:56where is it here? Is that you can run Android apps on here and so if you've got something going on on your phone and you want to finish it up, you can run the Android app and get it going. I'm not finding it here, but what do you think of this? Do you think it'll work? So a couple of thoughts. I mean, one, why why is the action to shake your mouse? Like why is that what initially initiates that that process? I mean, I think with my own mouse usage habits, like I especially if I've drank a lot of coffee, like I'm moving that mouse all

20:25over the place, right? That's right. So you're right. There's there's got to be some sort of calibration to where it doesn't just trigger every time you move the mouse rapidly. Like why would they not just put in a button that you could press Right. to to activate that feature. And then the other thought or observation is it's interesting that this is Google's hardware play when it comes to AI. Like we've seen that Chromebooks, at least from what I understand, have not been that commercially successful. I think the sentiment around them, at least my sentiment, it when I think of a Chromebook, I think of the cheapest

20:56possible computer money can buy that is not very [snorts] powerful at all. Right? So it's like of all the plays they can make around hardware and all the resources they have, why this? I'm sure they're seeing something that we're not, but the at the end of the day, I mean, Google has so many different things going on. They have so many different AI products. They're I think they're spread so thin that nobody knows them for anything specific at this point, aside

21:22from search. And and certainly nothing when it comes to AI, so I don't know. We'll just have to see how this plays out. And we'll also have to see how the other model companies react in terms of what hardware plays are they going to make? Like we've seen what Meta is doing with Ray-Bans. I've heard rumors of an Open AI smartphone. Uh I don't think I've heard any rumors around Anthropic in hardware, but we'll just have to see

21:45where things go. These people are a lot smarter than we are, so I'm sure they know They're they're doing it for a reason. Google has always been so freaking confusing. Too many things competing with each other, adjacent to each other, just versus Open AI, they're saying, "Look, you got ChatGPT." Now they're switching the branding to code to Codex. It's still one product essentially, and I think as they start to shift to both, it becomes confusing. Claude has one, Perplexity basically one, and then

22:12everything else is a feature within it. I'd like to see this work. I don't know that it will. It's interesting to to wait and have it come out. Um for now, I'm going to use my Mac and I'm going to experiment with products like Clicky, which is way I think more interesting and more powerful. All right, and there's the news. We will have a report below this for anyone who wants to see it all and read it with all with all the links that we've shared,

22:34and uh give us feedback. We'd like to see what you think of this. Yep, we'll be back. Thanks.