Billions, Browsers & What AI Will Do To SaaS + Video Maker MCP Preview - EP99.16-BILLIES

So Chris, this week my Dario pendant is back on and that's because Anthropic raised 13 billy at 183 billy. We're talking billies here. Post money valuation. The investment was led by Iconic Capital and they say it will help them expand capacity, which let's be honest, they do need help expanding capacity, improve model capabilities, and most importantly, safety research. Amazing. Imagine just that much money landing in your bank account and be like, all right, we need to spend all of this immediately. And they actually have enough stuff to spend it on. Yeah, I think you'd be talking like many variants of sparkling water in that office, like weird flavors that you've never even thought possible in the fridges because you've got to find a way to deploy the 13 billy. It wouldn't work for me because I would see everything in relation to how much money I have. And so everything would seem cheap, like literally everything of any nature would seem so cheap that we must buy it immediately. It's an incredible amount of money. And I think a testament in a way to how Anthropic really just quietly chips away. I think I read that they may have raised now more money than OpenAI potentially. Don't fact check me on that. It's probably wrong. but at least close and they're not they're not in the main zeitgeist right like they're i think they're making inroads but ultimately still most people think of chat gbt as ai and so anthropics just been quietly chugging along building models for probably the best use case arguably code and then focusing on enterprise use cases more and more and that's leading to obviously really good result and a pretty strong revenue run rate. But I mean, to justify 183, Billy, maybe that will look cheap in the future. Who knows? I also have to have a dimension of Sam Bankman Freed, the poor guy who's sitting in jail when his investors or people, sorry, his victims got paid out 130% of their money in restitution. So they made a massive profit more than you could get on the stock market or whatever, and his anthropic investment just keeps going up. The guy is and was a genius, and he's being persecuted unfairly. Free, what's his name? Sam Beckman free. Free SBA. We should start selling free SBA. I heard that they had to sell the stock, so now it would be worth, like, I think $1.3 billion or $13 billion. No, $13 billion. Wonder who they sold it to, huh? I'm just spouting off facts that are totally wrong at the start of the show, but whatever, it's an average show. But the other sort of billy news, while we're just covering AI pendants, AI billies, was that OpenAI have expanded the sale of employee exit liquidity. That means so the early employees that had stock or employees that have stock in the business that want to sell their stock or billies, you know, can get out in these exit liquidity opportunities. So they sell to secondary investors, essentially the employee's stock. Now, they've had to expand it because so many employees, oh wait, either so many employees want to sell or there's so much demand for the stock, either or, not sure. but open ai employees are now selling more than 10 billion and billion worth of stock to investors i mean at what point are you just not motivated anymore to continue to work like if there was if i gave you a billion dollars today and then i'm like get back to work on making chats finally fork in chat gbt are you really like are you really going to be like i'm so motivated by our agi mission um i better get back to making forking chats work there's probably some of them who are that are really dedicated to the cause. But I tell you what, it's really, really going to change your frame of mind when you've got $100 million in your bank account or something like that that you've earned very quickly. Yeah, I'm just not sure that you stay that motivated and you sort of – or do you just get greedy for more? You're like, how many more sparkling – like I could have my own sparkling water fridge with 72 variants in my house now. Fridge? You could buy the factory, Mike. Yeah. Have your own. But yeah, no, I agree. I mean, it's smart taking money off the table like that while the investor dollars are around. Like if they're willing to give you the money and they're going to let you cash out a little bit, I think it's very prudent of the employees there to do so. There's nothing wrong with what they're doing. But, you know, as an investor, I'd be thinking, you know, is this really the best way to get these guys to number one? It probably isn't. So while we're talking about throwing around billies, you know, obviously, Australians much more average, not really throwing around billies, but hundreds of millions, that's still acceptable today. Australia's 172nd world's favorite billionaire, Mike Cannon Brooks, put together this video after acquiring the browser company famous for making the ARC browser, which then later pivoted into the AI browser DIA, which I don't think anyone really liked that much. But here is the video and I want to talk through the video. I do think it's worth listening to as painful as it might be for some because there's some interesting points in the video. I'll roll the video. Hi, everyone. Mike here. Today, I want to tell you about a bold new chapter for Atlassian. We've entered into an agreement to acquire the browser company of New York, the team behind the incredible Arc and DIA browsers. Together, we're going to reimagine the browser for knowledge work in the AI era. This is a huge opportunity for both Atlassian and the browser company. And here's why. We have the potential to change the way 1 billion knowledge workers use AI to get work done in their browser. You see, there's a problem that every knowledge worker knows all too well. What do you notice about these tabs? They're not all recipe sites or newspapers or friends' feeds. Overwhelmingly, these tabs are SaaS applications and documents because knowledge workers don't just use their browser to browse. They use their browser to work. So whether it's your email or calendar... All right, we've heard it up. We get the idea. so first of all props to this guy for not developing like a posh billionaire accent now he's rich he just sounds like oh you dumb and bloody gonna need these browsers to do your sass work you know like he could he could just as easily talk like that and sound like he's down at the pub i like that i admire the jab at the uh recipe sites and you know it might be ego driven but i can't help but wonder given our frequent mentions of recipe sites and other trivial uses of this stuff that it was a shout out to us. But, you know, I think that's just my ego. It could be a shout out. It could be a shout out. I think he, like, you know, so we've talked on the show quite a lot about the evolving interface of AI when it comes to everyday work and SaaS apps and things like that. So when we started using MCPs in Sim Theory specifically, I started to notice, like, I didn't log into certain applications anymore. Specifically for me, mostly like the data apps, where I would just ask them, like, tell me the latest SaaS metrics or tell me this, tell me that, the email and calendar stuff I use frequently. So I'm constantly getting data from it. And I would also argue once more MCPs evolve to a state where they're consistently useful, like say a Salesforce MCP in the workplace, I think you would find sales reps and even leaders in the business relying on the mcp protocol to access the data through the chat interface more and we've talked about so many times that this ai workspace is really going to become like the interface to all sass and then eventually you could imagine over time as the moles evolved just replacing huge subsections of the software stack in a business in the ai workspace like why have all the disparate tabs and apps when you don't even need them and they're really just processes on top of dumb databases that can be easily replaced by AI and more customized to that particular business. So I kind of wonder with this acquisition, because I was trying to figure out like, why are they doing this? And it seems like, you know, if you look at ChatGPT, like they're going to obviously release a browser. Complexity's released a browser. It's like Search Engine Wars V1, where everyone thought they needed to control the browser to win and maybe you do but Atlassian thinking from the business point of view I would argue like hey eventually you could build a far better version of Jira maybe more customized to your business you could build a lot of this stuff so you know do we get the zeitgeist of having like all of these users that love this browser and then focus on the productivity aspects to flog off more of our software like that that to me seems like they feel like they need to own the entry point to where you access AI potentially. Do you see it that way? Yeah, it's a little confusing for me because I just keep thinking like what advantages do you have if you control the browser versus say a browser plugin versus some sort of computer browser automation where you're just running existing browsers, you know, in the cloud using AI or genuine computer use like we obviously are motivated by. I'm just thinking, what is it about controlling the browser that actually helps you? Can you run a model within the browser that has more direct access to the DOM or can control it in that way? I just don't see what the angle is in terms of what major advantage advantage do you get that's going to take people in their workplaces to switch to your browser? Keeping in mind, the very market they're targeting, these corporates, have internal policies that take a long time to develop, where they're like, you know, we use Microsoft Edge with these features disabled at the corporate level, and it goes through all that Microsoft stuff that controls what people can do with their browsers, how the profiles are managed, and all that sort of stuff. Same on the Google side. It's a very tough market to crack and it's fiercely protected in terms of people using it. And I would argue the only people using alternate browsers are people who are on the cutting edge and just wanting to try all the latest stuff anyway. So presumably there is some cohesion with their knowledge of AI because if you're using some weirdo browser, you're probably also aware of the latest AI stuff. but getting that to the level where he's talking about a billion people using it, they're really going to have to put a lot of money into it to compete with the incumbents there. Yeah, and I kind of wonder if it's a classic Jeffrey Hinton still relevant play where it's like, hey, Atlassian has something to do with AI, we're still relevant in this AI era, and they're just buying a brand that is in the, as I said earlier, the zeitgeist of people, like it's out there um that like for good or bad it's got like that deer browser got a ton of hype ton of hate um people really like the arc browser i guess because of vertical vertical tabs we proved them wrong with sim theory going the other way uh but yeah i just i i don't get the play here because it's like you know is the right angle the browser like the penetration of trying to convince everyone to use the browser unless it truly is better at workday productivity where companies are scrambling to get access to it then i'm not so sure but you could kind of look at it as a trello type i mean you can't argue with how they did well with trello but i think the second point i want to make about it where i don't know if it's the right way to go is the change the the vastly changing nature of SaaS as we know it. So the MCP as a concept has taken hold. When you look around now, more and more companies are proudly announcing that they have MCP support for their system, which means that there's already a better interface in a lot of cases to operate these systems than browser automation, right? So the usefulness of controlling the browser will diminish as all the different SaaS applications actually support a better interface for AI, which is the MCP slash tool calling slash API integration level, right? So it's obviously less efficient for an AI to be clicking around a web page as if it's a human when it can actually just construct all the parameters it needs to do operations in that system directly. Like it just isn't necessary for it to do that. The reason we got excited about computer use and browser use is it's a universal interface and whether or not that application cooperates, you can control it. But a lot of them are cooperating. And I would argue that the ones that cooperate are getting more attention. Like look at the HubSpot MCP, for example. Like that really makes that marketing automation tool a lot more appealing because you have a way to control it like right now. And so I just feel like we don't know with regards to SaaS what the actual interface is going to be. And as we've discussed before, does it become a dynamically generated interface? Do the apps become redundant when your AI just sort of knows what those operations are and can just do it itself? So to think that, okay, right now everyone's using their browser to do all this stuff. I think we're already seeing that moving away from that and that changing. People are using their browser less to do things because they have more direct interfaces through the AI into things. I think also the intelligence of if you've got like seven SaaS apps open, right? And maybe this is the play, right? Like I'm logged into, I'm authenticated into seven SaaS applications like Salesforce, some HR app, whatever. So the browser can look into those apps, right? So Atlassian, therefore, can look into those apps, see what's going on and draw context into a model for your workday. So it's a way of them having a product layer on top of all the applications you're using each day. I mean, that's a security nightmare, obviously. Well, yeah, but you would have to assume like this browser eventually is like... Asking permission and stuff like that, yeah. And if it can, then it can surmise data from it. But like you said, what's the difference between just having live connections that don't have to rely on what you're seeing with MCPs into those apps? Like, why? Like, to extract the most value from AI, like, you don't really need to see, especially in a browser. I think arguably for applications on the computer, that's a different story. But for me, it's like, okay, I can already get all this data from the MCPs across my stack. and I think the value to a business is far greater connecting all those pieces together and the individual knowledge worker because the knowledge worker now is like I can just chat to this thing it can pull in relevant data from all these things whereas before quite frankly it's an integration nightmare like think about any like the main requests in teams at a company oh can we get an integration to this oh if we just had access to our internal data if we just had access to our knowledge base coupled with this. Like that's night and day, like the main things that are going on in organizations. It's always been an integration problem and story where you just can't collate or gather the data in a format to then go and like analyze it. But now it's like, just connect, connect, connect, connect, connect. And then the AI knows what to call, how to extract the right context. It can visualize it. It can write documents. It can write reports. Like to me, the unlock there and like the cost saving to businesses, like removing the integration bottlenecks internally. And now it's at a stage where like anyone in the team can really, will eventually connect in those disparate data sources. To me, that's the real value here and probably not, I don't know how many people are seeing that or not seeing it today, but I think that is the big unlock, which then leads you to go, So whatever interface I'm using to connect into these MCPs and use it, and that's my like AI workspace, to me, that becomes the primary entry point, not the browser. Yeah, because people are using those browser interfaces because that is what is available for the software. And cloud-based SaaS software became convenient because you can work from anywhere. Everyone's using the same system. It's a common entry point for the stuff you're doing. But now there's a new common entry point that's better. And as you say, there's a lot of situations now where we're seeing companies bringing their own proprietary data into the mix, where it can work with the other SaaS applications on a different level, like in a different layer. the last thing you want is like a web browser taking your internal information from some sort of internal web interface that it's able to navigate extracting that information then pasting it into some other system painstakingly by like finding the fields or whatever so again i just don't see what advantage you have by controlling the web browser when it's going to be such an enormous investment in terms of um you know being able to comply with what's needed to deploy that in a corporate environment and getting the market share and convincing people to switch and all that sort of stuff. It's just an absolute insane undertaking when you don't even really need to do it. Yeah, I mean, the only use cases that are slightly interesting, but it could be achieved through like a plugin in any browser is if you're looking at like five articles and you're like, hey, based on these terms, summarize them. Like, I'm just not sure how compelling those use cases are. Like, eventually, I think any, like the native computer will be able to do that so who cares um and i don't then the whole dear browser of it having an agent and being able to like do things like book a holiday a restaurant or fill in this form in in salesforce or whatever as you say like it the mcp protocol is better anyway and i don want an ai over my shoulder as a knowledge worker watching what i doing in applications through the day unless I need it like I like come on help me so I don want it watching me in that case or yeah what kind of surveillance does this lead to if you're an employee like you install your ARC or do your browser and this thing's like over your shoulder being like not very productive like can't go on YouTube like I could do with that to be honest yeah it needs to probably help me be more productive but anyway i just i don't know if that's gonna like people are gonna want that and then the other thing is like the automation component of like oh cool i can now get the browser and other tabs to go off and do tasks for me cool like but i'd happily outsource that to another machine i'm authentic maybe this is my biased view of it but to me i'm like go off and do that ai like i don't need to see it or like monitor it in a tab in my browser um i don't know i feel like i'm slightly contradictory there but um i don't know these guys are super smart and they've made good acquisitions in the past so there's probably something we're missing but it's just not exciting to me in the sense that i just don't see how it fits into the future of where this technology is going look if this gets dominant enterprise market share within the next 24 months i will get a sim theory hat blend it in a blender and slowly eat it on successive episodes of the show like i will eat my sim theory hat if i'm wrong on this that's how confident i am this will not get high adoption um okay that sounds good what's high what's give a metric well i i don't count their existing customer base because to me like them selling more subs like some sort of browser sub into their customer base or just do it just do what google does where they're like you know you have no way to opt out but we're adding gemini to everything and increasing your costs jira no longer works in chrome bummer yeah yeah i don't know like if i'm looking at it last year as a business i'm like you've got a software suite that's aging it's over complex it's deeply embedded though and we know how hard sas revenues is to like destroy like it's impossibly hard and so i don't think they're going to go anywhere or anything terrible is going to happen to them but i think they're probably unless they start making these kind of acquisitions a company in decline long term because the whole software market's going to radically change so to me this is sort of a still relevant particularly when someone someone can use something like create with code and and make their own trello in about 30 seconds yeah and like the thing is couple that soon or like some of these other services already have it with like databases authentication compliance built into an ai workspace like imagine chat gbt releases a sas app builder which they very well might at some point with databases connected authentication like all the compliance stuff sorted out and you can sort of vibe design internal systems or like even a system just for a project, like just a singular board. And then because the data in these apps that you build is connected to the AI mothership, all of that's then available to the corporate overlords anyway. Do you know what I mean? Like it's like centralizing the data. Personally, I think what we're going to see is higher levels of abstraction in the way people, information workers, let's call it what they called it, how they work. It's going to be goal setting. It's going to be at first micromanagement of agents and saying, okay, I need this agent to do this, this one to do this. Now you guys mash it up. But then that's going to go to the next level where it's like, okay, I'm going to set you some specific goals to complete. And then I think it'll get a higher level of abstraction where it's like, make revenue go up and then they set their own goals and go from there. So I don't think getting down to the nitty gritty of the actual mechanisms in which the tasks operate matters to people. I don't think they care. I think what they want to do is get to the point where the AI is more and more intuitive at knowing what is needed based on the context. And you are telling it, here's what I want to do. You're giving it feedback and that's getting to higher and higher levels where you've got a team and you're telling them general objectives and they're getting them into the specifics and getting it done. I don't see it being like, oh, this is great. Now I can run Salesforce, Jira, and whatever using an agent in my browser because people don't care about the individual SaaS tools. They're trying to accomplish something. Look at Help Scout, for example, for tickets, right? People don't want to answer tickets. They want to help their customers solve their problems, right? The goal isn't answering tickets and clicking the reply button and entering templates and stuff like that. The goal is actually solving the problems people have. Now, this is something that we've discussed many times. Like a lot of help tickets are, I need a refund. I need to know how to do this in the system, point me at the docs. I've got a bug. All of those things with the right combinations of tools and agency are going to be able to be solved eventually by AI agents. So thinking that you need a way to control the SaaS app, which is your help app, is the wrong way of looking at it it's like what is the actual general problem the ai is trying to solve and can it solve it completely or at least more and more of that this is where i think sass is going it's going to just be more granular to a particular org designing an agent with inputs being like email mcp maybe phone um like it's basically like inputs and then the sort of agentic loop that solves the problem or escalates in some form internally and then spits back a response so if you think about customer support in general like you just don't even need a platform or a you don't need anything it's just you build and design an agent and tweak it and then all of that data is available in the ai workspace mothership and so then you just access that data and like yeah and then then it really becomes like a permission management system it's like expose this to the customer you've got an ai agent that can basically do all these operations in the system you ask for what you want it'll say if it needs approval and then go off and seek that approval like there's a lot of steps that can be just completely eliminated from these processes because of the agents yeah i think you've got to think about these things more like just a super employee where it's like you know in an early business prior to having this sas software you would just give someone an email inbox right and people would email it and phone like you know a landline maybe and they would interact with the customers and to me it's sort of almost going back to that level with an agent where it's just like super human though and can still interact with those uh channels and then i think eventually you could see like agent to agent stuff where one agent goes to the other agent like oh hey i need some help in context for a problem i'm trying to like hook into your program or whatever so i i think this is where like ultimately sass gets disrupted i it's gonna take like a decade i don't think it's gonna be some like epic uh steamrolling to be honest like it just aren't like just dealing with large businesses frequently you can just see how slow they are to move and how hard it is to like unwind things or change things. I think it'll take a while, but eventually the cost benefits and time savings will be so contagious that if you don't do it or get ahead of it or move with the times, then, you know, like you're going to fail as a company or at least, you know, the gains are not going to be as great. Yeah. And I think a topic we're going to also talk about today is this idea of the AI information gatherer slash mcp producer inside a company and the role they're going to play in this future because i think that this is going to happen a lot quicker than people are thinking yeah i i i mean we can talk about it now if you want like i i think that the like it's sort of like before the shift happens it's really hard to imagine it i think that's the challenge we all have right now because you you have these philosophical discussions like this one where it it seems pretty straightforward how it all will play out like that there's missing like a lot of missing pieces along the way but ultimately you're like i feel like i know where this is going roughly and i think the challenge then is you sort of go back to the way you're working with this technology today which is we're really in the infancy of working with mcps i mean And we're only just landing on a protocol now or the latest implementation of MCPs, which seems slightly more reliable and robust today. And so I think it's really in its infancy of how to do this stuff. But in its infancy, these early adopters are starting to get pretty good benefit, I think, at least. And that's where the value of tapping in, like I said earlier, to all these disparate data systems as MCPs and then conjoining the data, which otherwise, if you think pre-MCP, how hard that would be to do, like you just never would have touched that project. Exactly. And one of the examples I wanted to give, I forget what system it was, but with our old system, we had a thing where people could ask for feature updates and requests for things to be added to the system. And there were like 300 of them and people would upvote them and we'd do the most popular ones and things like that. People are always wanting to modify the SaaS software to help them solve their problems for their company better. And they're like, if only we had this feature, we'd be able to do it. And I think what we're seeing now is those same people within the organizations are seeing with a bit of AI assistance, I'm actually able to make these features myself. Like I'm actually able to take our internal systems, turn it into an MCP, which I can then plug into an MCP client like Sim Theory or Cloud Desktop, and I can do it. Like I can actually do these things that were previously unachievable. Or we saw things like Salesforce where people can build these custom, what is it called? Like fireworks or something, apps on top of Salesforce that would help with business operations. And you'd pay some guy $200,000 a year and he'd sit there all day on a laptop making these little apps. Now, people inside a company can do that themselves with a combination of these AI technologies. So I think that what's happening is, unlike, say, the rise of the internet where it took a while for everyone to realize the massive network effects and other things that can come from it, in this case, people have immediately said, hang on, if I can get that information into here and that can then go into this other system that we always use, then I can actually solve this problem that I've had for ages for the company and perhaps replace a job or minimize the amount we're spending on this other system or retire that other system. from our company. And so personally, I'm seeing a lot of people coming along with these custom MCPs that solve their exact problems or even expand the capacity of what their business is able to do. And I just think this is where we're going to see a huge future is this idea of the information worker, not just being like Ken and Brooks says, the information worker who's just clicking around all day on SaaS apps and needs an agent to help them do that, but rather the information worker worker graduates into someone who's creating new stuff like i was going to say software but it's not really software but creating ai systems that make their business better or more efficient and make them a 10x worker like they're going to be the ones to create it they're not just consumers anymore people are active participants in this new world of software and i think it's like you can imagine the next iteration of that right like so it's like okay i've got a bunch of mcps that can either take action or gather context across my business whether that's like first party stuff like say a sales force that you already have or just a custom system you have in the business or a database so i think it's like a combination of all of it and then you think okay how much time might i spend working on like a board report or a weekly report in my department in the enterprise right and i've got to go and log into multiple systems collate all this data look for meaning in it try and understand of it and make sense of it put that report together and like email it out once a month once a week whatever it is or like in a reporting meeting but now once all that data is connected it's sort of like you can have let's be honest it's a cron job not an agent but a task that happens every Monday morning. And it goes and calls all the MCPs, gathers all the context, builds a beautiful report in the document editor, either sends you a draft first to review via email, and then you're like, yep, let's send it to everyone. And also, like, follows internal style guides, looks at previous reports, and adopts the format of that report. Like, things that bring it into context for their exact organization following their rules and adhering to things. So it's better than what they would have otherwise produced as well. But also, that then just destroys. I think companies like, you know, or maybe not in terms of data storage, but you look at like another big company like Snowflake, where really, let's be honest, there's just like an overpriced cloud SQL. A messy database. Yeah, an overpriced database in the cloud that people got conned into buying. Anyway, oops. But, you know. It's a lake. Yeah, a lake. It was a lake, then it was a cloud, now it's like... Think about how much data you could store in a lake. Heaps. A lot of data in a lake, yeah. Yeah, and then you just have an army of pigeons transporting that data around. I think they started putting server farms in cold lakes too. That was a trial at one point. Microsoft did it. They had shipping containers that would sink into the ocean or some bullshit, as if that works. Yeah, so... But I guess what I'm saying is, like going back to that earlier conversation about the disruption of sass and all the different applications you can you can just start to get closer i mean obviously you still need an underlying database but what's to stop you just hosting that yourself like any database that the ai can interact with at some point in the future and you can interact with um and then that sort of is the roll-up of these like sort of call the mini like custom enterprise apps that you build in a workspace plus like connections into other things and internal systems. And then that task layer automation where it's like, do the report, go and look for problems in this area, like analyze my team's performance and send me a report once a week based on what they're doing in Arc browser, like things like that. Yeah, yeah, exactly. I think a lot of that busy work and that work that people do, like creating slides for like a monthly sales report or whatever, or daily updates, like all those things. Like they're boring and they're trivial. But if you can truly get rid of them. I mean, just think of the cost reduction. Like I would be looking at my SaaS spend. I'd be looking at my software spend, my database spend, going down the line items and going, which of these can be fully replaced by an MCP? because there's going to be stuff on there increasingly that can be just completely swapped out for an abstract AI interface that does all the same stuff, but you don't have to pay some company a ton of money to do it. You hear this argument, though, from the box CEO and people, like, oh, but SaaS is so many processes and compliance things and all this stuff. And I think he's right in the short term, 100% correct. and there's like logic built into these apps that is just very difficult to copy and also a lot of the end users will want to like copying them I guess like 80 to 90% of the work is really easy now with AI I think to copy a SaaS app but it's that last like 10 to 15% which is such a grind that it's still quite difficult but I'm just thinking imagine like whether it's Claude or ChatGPT or SimTheory or Atlassian's Arc Browser or whatever your AI workspace interface is, you can only imagine a point where you're tapping into like authentication and the compliance structure and like all this stuff. And you've got like this sandbox where you can build those custom applications that do start to take away chunks of that SaaS span. Like Trello is the easiest example I can think of because we know full and well, if you put a good database structure and some user roles and permissions, and you built it into something like Create With Code today, like right now, you could 1,000% replace that software. Like no brainer. You talk about OAuth, but the MCP protocol already has OAuth 2.1 discovered OAuth and more and more of the MCPs are supporting it. So you literally plug and play, you click, you get the OAuth. Do you give permission to access this data to this thing? you can sign up for an account or log into an existing account you're authenticated your mcps away you've got world-class auth like oauth is great the token exchange token refresh all that sort of stuff um out of the box no one has to do any work at all it's in the libraries on both sides you just do that and it's going to be the same like you say for payments it's going to be the same for databases there's going to be these components that can just be bolted onto these things where the AI just straight up knows how to use them. And you don't even have to think about that when you're building a new app. So is the value, do you think, in the short term for people, the MCP layout? Because I think the value goes from building a SaaS application to just building an MCP in the sort of guise of SaaS. Like, you know, you sign up through the MCP for your account. It's a task management system. It has an output interface that's like Trello. And so you can just pay for that MCP instead and it's far cheaper and consumed through that layer? Or do you think the company itself goes and designs all this software? Because I'm just not sure people will do that in reality. Yeah, I don't know the answer to that. I think the real value add now is around the AI task management like the AI being able to run at set times the AI being able to run in procedural ways to accomplish specific things the AI being able to run in a smart agentic loop to accomplish a goal the goal setting, that kind of thing. It's going to be giving the AI the best shot at doing the job with the model and tools it has. I think the tools are important and what's going to really get us excited because of the interplay between the tools but ultimately it's going to come down to getting that orchestration right to give it its best chance of succeeding yeah and i like i don't by all means i think that struggle with chats like this is you do get philosophical and everything seems to fit together but i think that in the now the best skill someone can have is starting to I personally think play around with MCPs in the context of business to get access to that context and allow their team to take actions with those MCPs that would be helpful to them that aren't you know aren't going to be like security concerns or things like that like even if you think about a lot of the HR software today you know you can import and say like oh can you get me my latest payslip or can you put in PTO for these dates or can you like there's so many things that it can help with in that regard or for a manager it could be like show me a mini calendar ui of when all my staff are taking away this month if you're looking at scheduling or something like that so it's just so many little use cases like that that i think are pretty damn helpful today yeah and a good example we had from a real uh customer during the week was this idea that okay you've got retrieval augmented generation on knowledge bases and that's kind of cool for looking up information. But sometimes that's not accurate enough to get the job done. You don't want the AI working on summaries of, say, your internal corporate rules. You need hard and fast context that is like, no, this is the way to do it. So building an internal MCP that actually enforces those policies when looking up information from the company database as a custom MCP makes a whole ton of sense. It's like you now have control over that as a filter on retrieving that context for your information workers within your company. And you're not just relying on the rag system du jour in whatever system you're using. So I can see the benefits, like the vast benefits of that level of control and curation by companies in building their own internal MCPs. And I think This is something that I'll probably do some separate videos on, on good, sorry, best practices on how to set these up. So, yeah, I think there's a lot to, there's a lot to come on this. I think that this to me was like the first step maybe of an awakening around the larger sass orgs, like with Atlassian being one of them, starting to awaken and realize that there is a future in which they're threatened by you know either mcp or potentially the interface of the browser because if you think about open ai like i'm almost certain they're going to release a browser i think it's been rumored like a thousand times and then i saw this tweet today it's from a verge senior ai reporters that take it as a creative soul they never seem to be right about anything but uh it says open ai is now open ai is now working on a linkedin competitor the open ai jobs platform is a hiring platform that will use AI to help find the perfect matches between what companies and governments need and what workers can offer per a new blog post from its CEO of applications. This was a new buyer or like a new someone who joined the business to be CEO of basically SaaS applications. So they're clearly taking, like they've put someone in charge as CEO of SaaS applications at OpenAI and he's going to go off and build a LinkedIn competitor. It's reminiscent of the early days of Google when they just tried a bunch of stuff and then just shut down things at random that didn't work. And then, you know, I think all the good stuff came through acquisition outside of Gmail, right? But I think the interesting thing here is OpenAI is now going, well, how do we extract maximum shareholder value on top of the AI system? Do you think they care about shareholder value or do they think they're worried about raising the next $10 billion and cashing out more of their stock. Because it seems to me like launching a web browser is a great way to get a higher valuation and therefore cash out more of my stock. But to take another angle on that, so Altman's famously gone around being like, I just don't know what we're going to do when all the jobs are gone next year or whatever. And then also we need WorldCoin and Universal Basic Income. But then, like, here we are, and they're building a jobs platform. We still want the real coins, actually. Like, OpenAI jobs platform, don't you see the joke? Like, it's going to replace all jobs, but we need to build another jobs platform. Like, what on earth? Yeah, you're right. I didn't think of that vast, vast hypocrisy. Like, it's pretty strange, right? Like, oh, don't worry, AI is going to, like, automate a bunch of stuff and reduce. But we're going to build a web browser, the technology that really came out about 30-something years ago, more, 40 years ago. Yeah, it's weird. And we're going to do a jobs platform, one of the first, like, websites to ever exist. Do you reckon those hype boys that were, like, selling all their, like, assets and stuff and being like, we've probably only got two years left, I'm going to enjoy it before the robot? Like, remember in the early days, like that, like, I forget that guy that was getting around eating donuts, telling everyone we're going to die. And then Hinton's out there like, you know, my chatbot's going to kill you all. I had to quit Google to warn you guys about what's going to happen. They're going to release Hermes 4 and it's going to take over the world. Yeah. And then we're getting a jobs platform from OpenAI. Like, could you have ever seen this like three years ago? like we'd be talking about a LinkedIn competitor from OpenAI. I mean, I've said it from the beginning. They are not a software company, like as in like a web software company. They don't get it. They don't have experience in that front, and they're just trying to throw stuff up, like you say, and see what sticks. I just don't think they've been able to commercialize on that side the way they want. I mean, chat GBT, though, is like arguably the most fastest growing consumer web app of all time. I'm not sure I agree. Because it's most people's most direct access to the model. I mean, it's pretty basic in terms of what it does. I'm not saying it's bad or anything. I'm just saying that they're not... It's like Google. They've had a few hits, but really their search engine core and their ads engine are what made them. And everything else has been done through acquisitions or some internal side project from someone. They've never deliberately set out to do something big and succeeded. Yeah, I think you're right. Like outside of ChurgyB, a lot of the things they've launched have failed. But I think bringing in a CEO of applications is interesting. I mean, you could argue it's like, are they spreading themselves too thin? Like this guy's going to go off and build SaaS apps and then you've got the core business. How does this all fit together? Like, do they have a clear vision? I think they are just trying a bunch of things and seeing what they do. I mean, as a customer of theirs, how about this? Make GPT-5 faster. Like that'd be really nice. Like if you just make it a bit faster, it's a great model. It's too slow. Make it faster. and then we'll all pay you the API money every month. Like, what's wrong with that? You're making a ton of money from that. All right. So, moving on. We were talking about the value of MCPs and then this idea of doing real work. And I think we've talked on the show many times about this idea of, like, all the pieces are there. You can just, like, do stuff and stitch it together. And I always feel like when we say stuff like that, people are probably like, oh, we're just, like, AI hype boys saying this stuff. but you know it's not like it's not necessarily the here and now and yada yada anyway so outside of the success of the podcast maker mcp uh that i launched last week or the week before um it's pretty good i've been listening while i cycle to customize podcasts with the two hosts i like the girls combination the best they've got a good good sort of back and forth um and i've just been learning about stuff like picking a topic saying make me like if i'm going for like a 30 or 40 minute ride i'll try and make it as long as i possibly can to fill the ride i've actually been enjoying it it it still would be better if it was a bit more like like like our podcast like just a bit rough and you know you feel like they should make mistakes tell people actively not to listen to the podcast facts should not true things like that would be more interesting um but anyway i saw someone on x had uh joined a bunch of models together in this like insanely complex workflow with this documentary um in the roman empire and they were like oh you know if you join like these 300 apis together you can now make like this custom video it only took like you know 38 minutes in processing time and cost 400 and cost yeah like 400 so i was like is there like can i get this price down to a point where yeah okay maybe it's not that viable today but like could it be viable like if i'm a teacher and i'm trying to teach a science lesson or i'm trying to teach a pretty hard concept and i have a studying yeah or studying and i have a methodology of how i want to learn that what if i could build a customized video like a youtube video or documentary on a particular topic with really good b-roll really good narration captions if i wanted and then i was like well there's a lot of good storytelling on apps like tiktok and the instagram video reels that people just love what if you could also make them and uh and so i was like we have all the components we've got and you thought there must be a better way there must be a better way my pitch is so lame but i was like podcast app got that got a lot of video models now that are great got a lot of image to video models that are great for like character referencing uh we also have uh background music with suno so you could create your own like customizable background track for the video so i created which i'm going to put in the mcp store next week it's called documentary maker i might change the name to vid maker or something a bit more catchy i think documentary maker is a bit lame given it can do other things um the other thing that i have also figured out i can do is the avatar based uh tiktok videos where it's like a girl speaking to camera uh saying whatever the hell you want um with captions like the cool fancy tiktok style anyway instead of just talking about it for all the listeners i'll show you uh so my son is obsessed with robin hood don't ask why um just move on from that uh but so basically what the documentary maker can do is the following you can be having a conversation with your assistant and then say like turn this into a documentary just like the podcast maker um it can make them uh 30 seconds one minute two minutes five minutes ten minutes so i like you can go higher but the cost is just insane um i am working on ways of reducing the cost so or just use your company credit card yeah or just use the company credit card um so it does you have a male and female narrator but i'm going to add the option so people can train their own voice so they can be the narrator if they want to start their own like youtube or tiktok channel using this tool um you can it does burn in captions like tiktok style with the car either the karaoke theme or this like the sort of expansion theme where they like get bigger and it's really catchy um it creates a customized soundtrack for the video uh perfectly cut to length in the background it stitches all the clips together it puts nice soft fades in i can't believe it worked i was in the shower thinking i wonder if you stitch all these things together if this would work and i didn't think it would work i my conclusion was there's no way. Turns out, is way. Is way. What's funny is if you'd done this 10 years ago, this would literally be like a sort of rags to riches story. It's like, I just was in the shower thinking, like, what if I could make just videos with captions and audio and stuff out of nothing, out of thin air and it would research the information itself. People would be like, you must be crazy, Mike. So here's all you do. My example, the video player is not showing and I'm still working on others in the background that haven't completed, sadly. I've got one that's almost ready and I'm desperately hoping it's ready in time um so so i just said create a 30 second documentary about jeffrey hinton explaining why he's still relevant to ai make it 30 seconds long with captions now i do know the caption system i broke right before we went to air and i haven't had a chance to fix so it's going to look bad but whatever meet jeffrey hinton the godfather of ai his pioneering work on neural networks and back propagation laid the foundation for today's ai revolution from chat JPT to image recognition. His research is the hidden engine. Now Hinton is a leading voice on AI's risks. After leaving Google to speak freely, he champions responsible development. His ongoing research and mentorship make him an indispensable force shaping AI's future. Okay, so that's a pretty cat like example, right? And admittedly, I've got to work on the prompts around the video. For some reason, the TikTok style ones, like the video clips aren't terribly good, but pretty amazing right and then like this one I did on understanding space time this is one of my proudest achievements from it imagine the universe not as empty space but as a vast flexible fabric this is space time where space and time are interwoven as Einstein revealed massive objects like stars and planets don't just sit in space they warp it so I think the implications of this Okay, flat earther I made you a documentary As if they know though, like seriously Wait, I've got a flat earth one Alright, good Ever wondered why ships disappear hull first over the horizon? Or why time zones exist? These aren't tricks Even the curved shadow Earth casts on the moon During an eclipse points to one Simple fact Our world is a sphere Allegedly do you want to know something funny about this video it tripped the the safety on on uh the video generation i don't want you to know the truth that's why yeah it tripped safety now tell me the conspiracy on that but anyway i've been creating like a ton of these videos right so here's the robin hood one i alluded to earlier long ago in a big green place called sherwood forest lived a brave hero named robin hood he saw that greedy rich people had too much while many poor families didn't have enough food. Anyway, you get the idea. But pretty crazy, right? It's like a full cartoon animation. The B-roll is perfectly synced. The captions perfectly sync to the audio. I mean, come on. It's amazing. And like, think about that. I think the important thing to think about it is it's not just the ability to create those. It's the ability to create them from any context you can generate in the Sim Theory environment. So it's like, it could be your emails. It could be your Monday morning briefing in your company. Yeah. Like go through your company financials and have a funny little video with captions and audio and music and all that sort of stuff. Like it could be so many different things. And like you say, in the education context, it's absolutely wonderful because you can make 10 of them, queue them up on your kid's phone, and then they can flick through it like it's social media. Yeah. But they're learning. Yeah, that's what I mean. TikTok learning. And the other thing that I think is cool with it is imagine, think about the market, right? Which we won't capitalize on, obviously, but someone else should for corporate videos, right? So everyone has corporate videos for training, workplace safety, things like that. They're all outdated and terrible, not very engaging and no one wants to watch them, right? You could take the existing video, upload it into a session and just be like, recreate this video but make it more interesting for like a millennial slash gen x tiktok audience and it will do it it will transcribe the video it will figure out what's in it yeah and then it will go and make a new video uploading existing assets would work just as well as anything else the other interesting idea i was exploring because i was looking for i was looking for examples of why for corporates having custom internal mcps would be valuable now think about this concept, imagine an AI that was aware of all of your internal systems and it could use Firecrawl or something like that to get the docs for each of those systems. It could actually access those systems internally. What it could do then is basically produce custom onboarding materials for a new starter in your organization. And with your documentary maker, you could actually go, here's your personalized introductory video that's going to take you through all our tools, how to use it, how you specifically log in and all this sort of stuff. Like there's so many possibilities in that area when you combine the different tools together. I also think just the way different people learn, like right now, I often think with my own kids, like I don't really want them going off and watching random YouTube videos, especially with the sponsorship stuff where they're watching a video about like, you know, space for kids. and then it's like a toy ad in the middle. Oh, yeah. Like, even if you pay for it, right? You end up in a supermarket buying, like, some random Mr. Beast bar or people are paying $27 for a drink of chemicals with that prime drink and stuff. Like, that marketing works great on kids. Yeah, and so it's like, I can make a video exactly how I want it to educate them on a topic. And obviously, you've got to assume this will get better and faster. Like, I know it will because I already, having built the prototype, them now like okay this weekend i can really get it humming like i i think by the time i launch it next week i hate the logic complaint during the podcast but my emu war video timed out and it was going to be amazing yeah well i gonna fix that i gonna fix that tonight that my plan i i i realized i can do things much faster and concurrently and but anyway i just wanted to get a prototype ready for the show but if you listening and you have some use cases you want this to work for and you join our discord or put them in the comments below if you're on youtube i'm curious like what kind of use cases you see for this kind of video gen now i know what a lot of people are thinking like what's the what's the cost um and the cost is we're probably going to lose a ton of money from this feature like all of them but uh the the generation 720p 16 by 9 uh for a about a three to five minute video i think is around six seven bucks now that sounds like a lot six seven have you heard this new kids joke thing you just said six seven that's going to trigger some kids i don't think i think our audience is like over 55s um it's basically a retirement home this podcast um but the the i think that's cheap like i know a lot of uh people in the community be like oh i can't afford that especially if it doesn't get it right the first go but think about how much it is to produce a training video internally at a business or if you wanted to start it even just like a tiktok channel with mini documentaries in your own voice with captions with b-roll like this is this is a pretty common way people make money online right they pick a topic they know a lot about they research it they produce many documentaries you could now do this and eventually intend automate an assistant with task management and the skill builder coming soon to to go and like publish this like every day at a random time a new documentary plus research like i actually think you could make a ton of money from this i had an idea of how to make it cheaper someone could actually sponsor the videos so if you use sponsored mode they can whack an ad right at the end so it's like your normal video then it's like drink pepsi yeah and we could have like a pepsi i'm sure yeah it's like it's half the price or yeah half the price or quarter of the price and someone shoves their promo we must have some rich companies who listen to this podcast sponsor the videos and we'll chuck your out on the end throw your ad on the free version yeah yeah do you want to see one more yes in the modern pursuit so i just before i play this a few caveats um so caveat one is i hadn't optimized the prompt at this point so it looks mental and like at one point the person in the video is drinking water and it like it's actually comedic how bad it is is this the one where they swallow an entire lemon and a glass. I fixed that. I've tuned that out of the prom, so we won't do dumb stuff like that. But this, in certain shots, I used ideogram character reference on my wife's face. I know no one watching knows what she looks like. But I would say this shot I have up is pretty close. Would you agree? Yeah, absolutely. I'm not going to name names, but it looks like your wife. Okay, so looks like her. So I use character reference. So this is the other crazy thing about the video documentary maker. If you upload photos, you can say, use these characters as references in the video. And it will. And it works. So you can be in the video. Now, it's not the best. It's not perfect. But check this out. This switch, primarily driven by a drop in insulin levels, is the cornerstone of how methods like the popular 16-8 intermittent fasting or more intensive 24-hour fasts can initiate significant weight loss. By simply changing when we eat, we take a thing. As the body enters a prolonged fasting state, it undergoes remarkable transformations, initiating ketosis where fat becomes the primary tool and autophagy, a cellular cleanup process that removes damaged components. This sounds like an American pharmaceutical ad. Glass turned into a pen in that one. Yeah. So anyway, I've tuned a bunch of that stuff out. The way I tuned it out for those interested was to say, like, I know where the video models are weak. So I'm like, don't do this. Don't ever do a shot of this, like, and so on. You did just give me an idea, though, of other ways this could be useful. Think of, like, doctor's surgeries and other places that need to disseminate information around different, you know, either topical things like, you know, vaccinations or when COVID was around and they've got to make little informational videos like this kind of thing would be amazing for that they can have them on their website they can have them on terminals in their office they can email them out like there's a lot of um value for businesses who need to to produce little info information videos being able to do this and in those cases like you say the price is totally and utterly justified paying some video company to do explainer vids think about we alone you and i have probably spent thirty thousand dollars or more i didn't even think of explainer videos right oh that's the one thing i'll say it takes a script so you can give it the exact script and say just make this explainer video so you could just write the script yourself it doesn't have to be ai generated didn't we once get sued for refusing to pay like 20 grand for an explainer video or something like that because they they didn't do anyway let's not get into all the galleries yeah like but what i'm saying is there is real yeah what do you mean mike i can't talk about it anymore or name names or the company they're all pricks but anyway the point is that people including us were paying real serious money to produce explainer videos and now you can just do it like that is pretty incredible it's so not to keep pimping this too much but the average corporate video in 2025 costs between 1500 to seven thousand dollars a minute so my seven dollars 60 per like five minutes or whatever is not looking too bad that's a bargain drink pepsi drink pepsi pepsi if you're listening call us i've probably picked a bad one there yeah i feel like no one from a software and companies listening, if you are, comment below. Yeah. Sun-kissed. What else is that for Stito? All right. Do you want to talk about horrific job loss for young people or just carry on with our day? Yeah, no, sure. Get into that one. So Stanford, you might know that organization. They're a university in the U.S., released this paper. It's called Canaries in the Coal Mine, six facts about the recent employment effects of artificial intelligence um especially in the corporate video making market that it's it's gone the way of the website but anyway so this this this paper essentially says that it shows that there's been a clear 13 relative decline in employment for workers aged 22 to 25 in ai exposed jobs specifically junior software developers in customer service since late 2022. You could argue, I would counter argue, maybe it's the economy, but anyway. My hot take gut instinct, zero research. Hang on, before you give the hot takes, let's just go through the facts. So the study does make a critical distinction. Jobs where AI automates tasks are shrinking for young people, but jobs where AI augments human skills are stable, if not growing, because the perception is the output of those individuals has increased, which is pretty interesting. First time we've seen that in a study. The findings, so it says it's not just a tech bubble bursting. The findings hold true even when the authors remove all computer-related jobs and tech companies from the data. The effect is also present in non-remote jobs. So like a lot of those arguments people have been making, like I just did around the economy, flawed. The mystery of stable salary. why while employment is falling for this group their compensation wages are not the adjustment is happening through hiring freezes and layoffs not pay cuts so it's basically and we kind of alluded to this earlier with the gains in productivity people just aren't hiring as much as they used to they're not laying up that people aren't getting laid off it's just the people working are expected to now do more because of these tools yeah that that is totally my opinion as well. I think people have just hit pause on hiring in a lot of the jobs because they just don't know what's coming. For starters, they're getting more productivity, as you say. And it's just, it's not that people are just actively being fired because like the robot sitting in the corner now punching away doing their job. It's just that why would we hire someone to do that when I can do it by typing a prompt into chat GPT? So that makes total sense to me. The other point I was going to make but i think you've sort of invalidated already with the actual facts which is annoying is that if it was me and i wanted to lay off a bunch of people while saving face and not looking looking like an asshole company i would use ai as an excuse to lay off unproductive workers and say oh these 10 000 people we fired is because of ai but it's not really we just wanted to get rid of them yeah i i mean like mark benioff was kind of i mean i wouldn't call it bragging but kind of bragging about laying off 4 000 people as a result of his claimed implementation of ai he also said that they were able to reconnect with over um i don't know if i believe this data 100 million previously neglected customer leads with ai i think 100 million leads how many people like this is like a continent the universe yeah um that i thought it was a hallucination but i looked up that direct claim unless it's over the cost like customers as well like 100 million it seems like way too big but anyway so i think you are seeing gains and they're just coming through slowly but it doesn't surprise me as someone who like does typically hire a lot of people and is involved in this kind of thing with ai tools now i know certain things can be done faster whether that's an api integration or writing a blog post or whatever it is is part of the workday that needs doing you just instinctively know now being an ai first individual you're like that's easy you can do that in like five minutes you can do the research think of a senior front-end developer building react components the ai can one-shot them that would take days or weeks for uh even a senior developer to do by the time they tested it, put it in different scenarios and stuff like that. The AI can do it in 10 seconds. I'm not saying it's perfect and obviously there's more to software development than just smashing out components, but bloody hell, it's a pretty big impact. And I think this would be the same in lots of industries where it's like producing reports, writing, drafting letters, making PowerPoint presentations, like all that sort of stuff. the heavy labor, the time consuming bit is disappearing. And it's more about iterating with it or vibing it, as we say, to get to where you want to be. So I think that is a productivity gain. You are finishing more things and taking on more things because of it. Oh, yeah. There's no doubt that it's a real thing. Like I feel it. Like it just makes you question every new hire now you're like well okay can i just automate this with an like with some sort of like home built agent or can i um can i get someone that's already doing that job just to use ai tools to be a bit more productive like i know it doesn't always apply but i think it applies enough there's also those intermediate jobs like you think about you know you've got again i'm talking software because i know it but like you've got a graphics designer doing concepts you've got that graphics designer then handing it over to someone to turn that into markup as like a prototype and then you're handing it over to developers those first two steps can basically wholly be done by vibing it out now with various tools like you can design design it get it turned into markup and then hand it over to your team that's two highly paid jobs that if unless you had really specific requirements you could completely avoid hiring for and maybe time will tell and you don't end up doing it that way it's just more convenient having them but it sure is hell going to put downward pressure on hiring in roles like that and that's just a couple i'm thinking off of the top of my head yeah i i think though the positive i'll say about it for young people if there are any that are listening that are like oh this sucks i don't think it does because i think it leads to a lot of these jobs in this sort of customer service area that quite frankly should have been automated. And a lot of the productivity gains you see, like the positive sign in this is people are getting paid more on average and not actually being laid off where they are more productive from these tools, which leads me to believe that if you're an AI first worker and you train yourself coming out of school or if you have a university degree and you're native with all these tools and understand them and can be more productive with them, it's so desirable to an employer like oh they're going to come in and help me transform this organization or they're going to come in and just step up and be at the capability of yeah and enlighten the other employees and get them more productive as well and i think the best way to get the job is to demonstrate like hey here's how i've used ai like these are the things i've done on my own accord uh in my own time because i'm interested in that that i can then apply into the business and into the role and i do think that provided there are entry points to talk to these companies and try and get hired then there is still opportunity for those out there that are willing to sort of somewhat retrain or just natively train in these tools. Yeah. I've got one more video before we go. All right. I found it finally. So I did a character reference of myself with my aero, ridiculous aero road bike helmet. This will just make you laugh. How many grams is it? I'm sorry for listeners. Very few grams. I don't know. It's like, it couldn't, would it be 200? I think 200 grams. Very light helmet. But anyway, it ripped the helmet off my head for this video. Watch this. Mike, a local legend, made cycling his way of life in Newcastle. So you'll love how it turns everything into a documentary and makes it seem important. Yeah, it could also be talking about a new kind of milk. Anyway, you'll be able to pick your own voices by the time this goes live. But check this shot out. I think I'm going to put it on the cover for this episode. It's so ridiculous. Anyway, so I look forward to everyone's ridiculous documentaries next week, if you can afford it. All right, that'll do us. Any final thoughts from our ranting? We just ranted for like an hour, 13. I can't believe people listen to this shit. yeah i think last last week i i said that we might have an update on the workspace computer i know many people are getting excited about the reprise of that yeah so yeah i've just there's been a lot of stuff we're working on i haven't got to the point where i'm confident showing it but we are progressing and i think next week's the week we're gonna have an episode where we gotta stop committing what just yeah there will be a week when we show it off that's right i refuse to commit to deadlines but i am working on it that's my point all right thank you very much for listening if you do want to sign up to sim theory and maybe make your own ridiculous documentary or replace a corporate video uh you can use code still relevant it still works it's still relevant still relevant coupon code on simtheory.ai it also goes a long way to supporting everything we do and the show and also paying to keep our servers live to make ridiculous 200 worth of testing videos for this show like this all right it has been our pleasure we will see you next week goodbye Oh, Dario's counting digits, 13 billion in the pot Part-time money flowing, whether he likes it or not Steady wasn't thrilled about it, but the game don't stop When you're building eight, our empire's got a roof to top From one to five billion in eight months flat Revenue explosion, imagine that While the competition's watching, trying to catch up But we got the formula filling up our cup Billies in the bank, yeah we got them billies stacked up, no need to think. Billies in the bank, open A employees cashing out, filling up the tank. Open A employees got their hands on the trigger, 10.3 billion making figures way bigger. Secondary sales got them feeling like winners, but Bill Gurley asking why the insiders are quitters. Stock option turning into Lamborghinis and chains, while the company's burning through retouching campaigns. 1.5 million just to keep telling place, but they still in shares at a record-breaking pace. 500 billion valuation got them feeling high, but we're inside a scale. You gotta question why, are they losing focus on the mission at hand? Or just securing bags while the market's still great? Billings in the bank, Billy's in the bank. Yeah, we got them Billy stacked up, no need to thank. Billy's in the bank, Billy's in the bank. Open air, employees cashing out, filling up the tank. Daryl's got his billions, anthropics still on fire While open A's talent's got retirement to acquire Golf money, tech money, everybody's buying in But who's really winning when the cashing out begins? Plots generating millions, half a billion in code While Chad keep he taste users down a different road Both companies rising but the story's the same When you got billies in the bank, you're winning the game Billies in the bank, billies in the bank Yo, count billions, golf money in the tank Billies in the bank, billies in the bank Employees cashing out, but who's left to think? Our revolution, but the focus might be shifting When everyone's selling, who's really still lifting? Billies in the bank But at what cost when the mission gets blurred? Is the vision lost?