SED News: Restricted Models, IDE Wars, and the DeepMind Mafia

SED News is a monthly podcast from Software Engineering Daily where hosts Gregor Vand and Sean Falconer break down the biggest stories shaping software engineering, Silicon Valley, and the broader tech industry. In this episode, Gregor and Sean dig into the growing tension around restricted AI models, including Anthropic https://www.anthropic.com/‘s Fable being pulled from the Claude https://claude.ai/ platform days after launch. They explore what Sean calls “vibe regulations” and the risk foreign governments and enterprises face when a model they depend on can be cut off. They also cover the FT’s reporting on London’s “DeepMind mafia,” a vibe-coding clone controversy involving YC-backed Corgi and Papermark, SpaceX https://www.spacex.com/‘s acquisitions of Cursor and Mesh, and Anthropic’s launch of Claude Science. They also take on the latest round of the IDE wars, and explore who owns your dev toolchain, the vendor lock-in that now comes from context and memory rather than the model itself, and the widening cost gap between frontier tools and open weight models. As always, the episode wraps up with a few standout Hacker News threads. The post SED News: Restricted Models, IDE Wars, and the DeepMind Mafia https://softwareengineeringdaily.com/podcasts/sed-news-restricted-models-ide-wars-and-the-deepmind-mafia/ appeared first on Software Engineering Daily https://softwareengineeringdaily.com.

• 0:11Speaker 00

  Hello and welcome to SED News. This is the monthly format of SE Daily, where we look at some of the tech headlines. We dive into a deeper topic in the middle and then we pull out some of our favorite hacker news highlights. At the end, as usual, we've got myself, Gregor Vand, and with me as usual is Sean Faulkner. Hey Gregor, how are you? Hey, everyone out there. Yeah, not bad. It's been, we keep using this word busy, but yeah, it has been, I think, exceptionally busy for both of us. What's been keeping you busy over the last month, Sean? I was on a work trip internationally and I also dragged my family along on that.

• 0:52Speaker 00

  So that was good and also introduced certain complexities, I guess, along the way. And then I've been back home for a bit, but then I'm leaving this weekend to join you. Well, not specifically to join you, but I will be in Singapore where I hopefully will meet up. Yeah, that's exciting. Yeah, so you're your first time to Singapore, I believe. Yeah, I'm excited about that. Yeah, no, I'm always happy when people pass through. So yeah, we will catch up. Yeah, and big news in your land or your world as well with Superbase and their Series F. Yeah, exactly. Yeah, we, we announced our Series F, which yeah, was like 500 million at 10.5 billion post. Yeah, which was just a bit of a huge step, I guess, for us.

• 1:36Speaker 00

  And not something that we maybe expected quite as soon as we went from E to F, if you know what I mean. Yeah, well, how long have you been? It hasn't been that long for you, right? No, it's only been about nine months. Yeah, so clearly you're the catalyst for.

• 1:50Speaker 00

  Yeah, well, just saying since, since I signed that contract, which was actually longer than nine months ago, that yeah, that's gone from D to F. So I've brought some sort of good luck charm, maybe. Yeah, clearly correlation equals causation in this particular example. No, if any of my colleagues listen to this, absolutely not. I'm just a sort of cheerleader, you know, making sure. Sure. People attend the right meetings and all that kind of stuff. Mm hmm. So yeah, it's a really exciting time. If anyone is listening and interested, Subbase is still hiring. So yeah, do check out roles there as well. But it is that kind of startup. It's just like it is a rocket ship and you feel it, you know, when you're inside. So. No, that's fun. That's a fun place to be.

• 2:29Speaker 00

  Yeah, for sure. Well, yeah, let's get on to the headlines. So the first one up this week is Fable and Methos. So these are the sort of complementary models from Anthropic that I'm sure most listeners are sort of, well, Methos probably more maybe than Fable. Methos being the very cybersecurity focused model that was only available to certain companies and even maybe some certain governments, but very much restricted. And then about two weeks ago, Fable was released, which was their sort of consumer grade of this and. It's like Methos with extra guardrails.

• 3:12Speaker 00

  Correct. Yeah, exactly. Yeah. It was interesting. I was actually doing an episode with, I wouldn't spoil it, but I was doing that for someone for SE Daily that morning, who's in the security space. And so he had been tinkering with it like that night. And so, and he's saying, oh yeah, Fable actually does what it's kind of says in the tin, which is like, it can find, really found things that I wasn't expecting, but then equally it also has the guardrails. Like it really did sort of stop me going further than I like. It was, I was trying to deliberately push it to go further just to see what happens. And yes, it had the guardrails. So I was like, oh wow. Yeah. And then within about what, two days, Fable had been removed from the Claude platform.

• 3:52Speaker 00

  And this was based on security concerns. So what we're seeing here though, is that we've also got the same situation kind of playing out with GPT 5.6, where they're saying that it's going to be only rolled out to certain people. And you know, this is because again, it's quote, dangerous in the wrong hands. And I think we're just starting to get to a slightly strange place of what is going to be the, the go to market of these supposedly super powerful. Models now, like if they can only be used by certain people and and then EG a government can then say, hey, well, if South Korea Telecom, I think it was like, is using this thing, you can't, we've got to restrict it now or something, you know, it's it seems a very strange place to be.

• 4:35Speaker 00

  Yeah, I mean, I was in Europe when this came up. And of course, I got a lot of questions from, you know, customers I was interacting with and things like that about what I thought about it and opinion. And I think the, you know, the big concern from a lot of countries is that, hey, if I or government, foreign governments, or, you know, even companies is if I like adopt this model, and this is the thing I'm investing in. And then suddenly, like, a foreign government can just say, this model is no longer available. But that's a major impact to like those types of investments and decisions you're making. So it's kind of woken people up to that being a potential risk. And I think like one of the things I've been thinking about is like, does this end up encouraging

• 5:16Speaker 00

  more use of open weight models or more use of the models that are being developed within certain regions of the world, like, you know, Mistral in in Europe is a popular model. But does that mean, does, you know, I wouldn't be surprised if Mistral got a, you know, bump after that or something like that from European contingency. And how does this kind of affect the decision that people are making around investing in these types of models? Like, I don't think anybody necessarily has these answers, but I I think it calls into interesting questions. I think there's also the question of which sort of market Andreessen articulated was, which is like, you know, this supposedly totalitarian regime, we're referring to the, you know, anthropic and OpenAI.

• 5:58Speaker 00

  They're trying to open up the technology and then the supposedly democratic system, which is the US government, trying to restrict and control the technology. And I think that makes certainly certain people uncomfortable, like how much control should the government have over over software like this? I don't know, it creates a lot of, you know, big questions. And I think also, there's also the concern, of course, that people have around Chinese government sort of catching up in this stuff as well. And if we slow down innovation within the US, does that give essentially the model race to another country? Yeah, I think I think that piece specifically is is interesting where and there's I feel there's a few things at play here. It's like, does China have an incentive at the moment to restrict models or like now?

• 6:42Speaker 00

  Does it kind of know that its models still have like ways to go on on certain areas? And so, yeah, this is kind of the moment to like, strike while they are and so if you if you like, and keep them as as accessible as possible so that people try them out. I mean, I definitely feel a lot more anecdotal evidence of people trying out open weight models now and I'm being less less bothered, concerned about the fact that a lot of them do originate from from Chinese labs, et cetera. So, and yeah, and then then we, yeah, we've got this strange situation of democratic countries being the ones to restrict it, but on the basis of our technology is absolutely superior. It's like so capable that, you know, if if this gets in the wrong hands, then then we can't have that. So it's just a very

• 7:27Speaker 00

  strange place to be at the moment. Yeah, and I think that it's not clear to what they actually mean in terms of like safety standards. Like, I don't think there's a a lot of this stuff has been like clearly fully thought through. It's not defined somewhere. So it feels a lot more ad hoc and subjective than it's like some like official checkbox or it's only getting your PCI compliance or something like that where there's like something written down that you can actually follow to show that like, okay, I mean, the regulation. Nobody seems to know, including the administration themselves, of what this is. You know, we talk about vibe coding. This is like vibe regulations or something like equivalent to that. Yeah, I think that's that's like you heard it here first vibe regulations. Yeah, there we go.

• 8:11Speaker 00

  Yeah, I mean, like GPT five point six codenamed Sol, S O L, which is the the OpenAI model that has been like so called restricted. Apparently, you know, it introduces a max reasoning effort mode and an ultra mode that uses coordinated subagents to solve highly complex tasks. That sounds pretty normal to me. But then I think. Again, it's the cybersecurity angle that has kind of undone this one. So, you know, to assuage any fears of its powerful models being unsafe, OpenAI has said that Sol includes its most robust security stack yet. You know, it's this hardened against adversarial attacks and intentionally optimized to favor defensive cybersecurity work as opposed to offensive exploits. So I think this

• 8:56Speaker 00

  is like where they have to explain back to the government what this can and can't do. And I'm very curious to know who in the government is, I don't know, testing this thing to its max limits on any of that, rather than just saying, oh, this, this sounds dangerous. We've got to restrict it until, until what point in time? Like, yeah, like, what do you classify as, as like, safe to release again? Yeah, I feel like that's going to be a difficult conversation, just based on if you ever watch any of the history of when like Microsoft or Google or Meta has had to explain their technology to the government in like official settings. It's like clear that nobody understands, at least the US government understands like the basics of how this technology works. Like now

• 9:38Speaker 00

  we're talking about like these deep neural networks and things. I think it's going to be a tough conversation to articulate. But it's interesting too, to see the reaction from the two companies and Anthropic and OpenAI as well. Like Anthropic's quote was that they're pleased to see the progress. And maybe that's just a throwaway line or something. But OpenAI is like, we don't believe this kind of government access process should become the long-term default. So they're kind of at least publicly on the opposite side of that. And maybe that speaks to the culture within both companies as well. Yeah, absolutely. So moving on next. So thanks to TechCrunch had a bunch of points

• 10:16Speaker 00

  from the Fable and Mythos regulation piece. The FT Financial Times, yeah, had an interesting article this week on London's AI boom via the quote, deep mind mafia. So, you know, like when we talk about mafias, it's often, you know, people that were in a certain company at a certain time and then they sort of, you know, in a sort of heyday and then they disperse and go and do new things like the The famous one is the PayPal mafia, which has like Peter Thiel and Elon Musk, of course, and the name's escaping me, but the person who founded Affirm as well. Oh, Reid Hoffman? Not Reid Hoffman. He was also in the PayPal mafia. Yeah, there are, I mean, there's a lot of them, right?

• 11:01Speaker 00

  Yeah, so he obviously Reid Hoffman LinkedIn. But here we're talking about the DeepMind mafia, which is, you know, DeepMind was like one of the first research labs into into large language models, basically, and was acquired by Google. And that really kind of, Google were kind of sitting on this thing for many, many more years than people realize, basically. That's why it was always a bit surprising or strange that Gemini was kind of lagging behind people like OpenAI and Anthropic. You know, we've talked about it many times. It's caught up in various areas, but the fact that we're literally sitting on this technology and sort of got beaten to it by these other two. But yeah, London has kind of become a bit of a hotbed for having where the researchers are and sort of the deep

• 11:43Speaker 00

  research on this. So, you know, Demis Hassabis is the DeepMind founder who's kind of, I think, planted that seed. But the problem that London seems to have is just that, but no foundational models are actually like really built there and or run there, really. And so when, you know, say Anthropic has to like stop a model from being used. I think the temperature here is that in London, they're saying like, well, we have all the researchers, we have all the technology, we just don't have like the company or the infrastructure running it. So what happens when an American company pulls the plug on things is this is a bit embarrassing. Yeah, I mean, I think if you look at it, the DeepMind alumni have raised like $55

• 12:26Speaker 00

  billion globally, but only $5 billion of that has stayed within the UK. And none of the alumni are building any of this stuff there. So that's the big thing that the article's highlighting and highlighting some level of frustration is that the UK is creating all these brilliant people, but they're failing to capture the value. And there's probably a lot of things that go into that, too, just where data centers, where some of the technology innovations coming from, where the money. To raise has come from, and I think that that has gotten a lot more global. Like, there's a lot more companies that even raise from VCs in the Bay Area, but aren't necessarily headquartered in the Bay Area. Like, it doesn't have to be there. But there is certainly advantages to being in the US,

• 13:09Speaker 00

  and in particular, in the Bay Area for technology companies. So I can see why some of that has has moved there. But there's clearly this kind of like brain drain that's going on between the like brilliant people coming out of DeepMind and all of these universities in the UK. And then the UK not, you know, being set up for success to capture the value for whatever that might mean. Yeah, I mean, it's a sort of well-worn path. I'd say like where a lot of amazing technology research and like R and D, you know, happens, especially sort of in like the Cambridge area. But yeah, unfortunately, most of these people or companies, they just quote, follow the money and the money usually is in the US.

• 13:52Speaker 00

  Yeah, I mean, when Google was acquiring like DeepMind and also all the, basically, they had all the the greatest minds in AI in the world essentially working for them at one point. Like, probably, I don't know, 10, 15 years ago. They all kind of, all these companies that we talk about all the times, like their their founders at some point dipped their toe in the Google, were Googlers at some point. It's kind of incredible that they were able to bring all those people together. But I think that talent attracts talent. And then they have basically unlimited funds to compensate those people and have them come work on those things. Like, I think if you listen to, I watched a documentary on DeepMind and really they had this aspiration. Even though they had raised money, they didn't really have aspirations around like being like a commercial entity.

• 14:36Speaker 00

  They had this ambitious goal of building like AGI or solving AGI. And they weren't really like building products. They were trying to like solve this problem of like AGI. It's kind of very research-driven thing. And it's hard to continue as a private company indefinitely with that. So part of the motivation was that they could join forces with Google. And Google founders essentially gave them the promise of like, hey, like you can go and do this thing and we will worry about the commercialization of this. And you can just continue to do your amazing research. All the pioneering work they did around reinforcement learning that led like AlphaGo and so forth. So they got to kind of just focus. Focus on the mission and not have to worry about the commercial stuff and the money is always flowing in.

• 15:20Speaker 00

  And then on top of that, they also have suddenly access to these gigantic compute environment that Google has been building over the last 20 years. Yeah, and it's almost that environment that it probably did a lot for the R and D side, but it hampered like how they actually get this out as a product. Cause you've got this kind of like specific, I read a book kind of covering all these companies and yeah, like DeepMind, they still sat in a very separate office. And like they got to do all DeepMindy things while literally you have like over the road, the Google office. He's like, hey, well, we own those guys and we've got, we should be able to use the technology, but like, how do we use it or how do we sell it? And anyone in product management will know that that's like not how products get released fast. Yeah. Well, I think they've, they've taken steps to unify those two teams now, the Google Brain team and the DeepMind team.

• 16:05Speaker 00

  But yeah, and certainly I think what's happened to Google and the AI race is a deep motivator for aligning those teams and becoming more focused on like getting products out than just purely the research mission. I think the interesting thing too is that the article highlights that like the UK's own tech advisors, including people from AI labs and DeepMind had told the government that building like a British LLM would be a waste of, of, of taxpayer money. But now, given all the things we were talking about in the beginning, in terms of the US cutting off access to certain models. It's now reinvigorated essentially this attempt that there should be a UK AI model that they own.

• 16:46Speaker 00

  So that they don't have to be subject to like the whims of essentially a foreign government. Yeah, I find that sort of interesting because it's not like there's a US government LLM like that was funded by taxpayers' money. But yeah, I think, you know, it's, it's just a different way of looking. You know, the US is just this amazing sort of capitalist machine, right? And the UK. Maybe the UK is thinking that the key is to move fast with products. You want the government to be involved. That, that's really is that if you want to move quickly, involve the public. Yeah, definitely heard that one before. Yeah. So really, I mean, in my humble opinion, how.

• 17:24Speaker 00

  Grown up in the UK, it's just the UK just needs to like, if they were serious about something like this, they have to just get more risk averse. People have to be more okay with, you know, venture capital, et cetera, et cetera. That's a pretty well worn understanding, I think, but I think sort of a British, a British L L M from the government. I'm not sure that's like, that's the answer, but I can see, I can see why they're at least just talking about it. Yeah. So I think going on to our next news item, so this is back to a tech crunch. So yeah, there was another Y C backed company where source code stealing was sort of a claim that has been made.

• 18:05Speaker 00

  So this is something that's kind of happened before the company today is, this one is, is Corgi. So like it's Corgi is a, like a deal flow kind of backend platform. Like you're storing all sorts of like data for VCs and this, this kind of thing. But they got accused of replicating another platform called Papermark. And it was kind of interesting because the post that sort of highlighted this to the, to the world, it blew up because I think one of the founders, he shared a screenshot. I think this is the Papermark founder, shared a screenshot and he showed Corgi's product like side by side with Papermarks.

• 18:47Speaker 00

  And it was like virtually the same UI next to each other, but they in a very specific place. It wasn't like the front page, it wasn't like the main dashboard or something, but it was like in these certain areas where the language is very specific. And so really the arguments come down more to it's not code copying because it was vibe coded. And so Corgi is arguing, well, we just told it to build the thing. But they have admitted that they did tell it to basically replicate Papermark, which just raises this whole new concept of like, what is copyright infringement exactly? If it's not exactly the code, but it's the idea. I mean, way back in the day I studied IPO law, so I need to dredge up some of that knowledge. But like, I think it's just a very interesting concept. Yeah, I love the, we didn't steal the code, the AI happened to produce something identical.

• 19:33Speaker 00

  But yeah. I mean, like, if I went and said to an L M, like, I want to write a book. I want you to copy Harry Potter, but change the names and the setting to some degree like. In my mind, I'm clearly breaking some sort of like law, right? Like, it's a direct sort of copy. So I guess the difference is that the L M can actually inspect the writing of the book. Whereas the L M hasn't necessarily inspected the right, like, the code backend of the service. It's just copying to the best of its ability, the front end, and then assuming certain things in the backend would need to be, you know, built a certain way. But it's a, it's a complex, very super complicated.

• 20:14Speaker 00

  It's not like people haven't been copying what their competitors or similar products have been doing forever. It's just it takes more effort to do it than. Exactly. And with the effort, people probably also put some thought into like, maybe we shouldn't copy this like exactly pixel for pixel or. Or they did a better job of hiding the fact that they were copying it. Yeah, I think the, the effort thing is interesting. I mean, because, yeah, like. Intellectual property frameworks, yeah, they were, they were sort of designed for a world where we're copying required effort. And then if you look at, yeah, the incentives for someone, like, let's just take the obvious one, like Instagram, you know, copying stories off Snap. It's not that they sort of looked at that and said, it is really worth our time.

• 20:55Speaker 00

  Putting a bunch of engineers and literally taking that concept and basically planting it in our platform. And we believe deeply in that. And so, like, they must have also been ready to, like, defend that. But if that effort is removed, then the sort of incentive to copy is just like so high. Because it's like, well, we might as well copy it, see if it works. If it doesn't work, we can like trash it. Go spin up a vaguely similar but different version. But the kind of, are we really going to do this? Are we really going to like, put the engineers on this? It's going to take like a few months to crank this out. Like, it just changes the whole dynamic of like, anyone thinking about replicating somebody else's product.

• 21:34Speaker 00

  Yeah, and clearly it takes a lot more than just like copying the code or copying the interface to like have a successful product. But I do think it kind of calls into question. Like, what is the moat for companies now? Like traditionally, or like, I guess like an unrelated thing of it is like, AI can basically reproduce the look and feel, the functionality of pretty much any SaaS product in an afternoon without even touching the source code of the original product. You know, what exactly is the moat of the software company? And it has to be either some access to proprietary data that the competitors don't have access to.

• 22:10Speaker 00

  And then another big one is certainly around like go to market. And I think if you look at this, you know, like all the money that like Corgi has raised in succession, you know, maybe that's kind of what they're thinking is like, hey, we need, we can essentially grow really fast, corner this market, put the money behind the go to market, but like, it's not about the money necessarily going into building the software, the software we can build relatively cheaply other than the token costs. But we need essentially to go out and land grab all the potential users and that takes money.

• 22:44Speaker 00

  Yeah. And speaking of money, I think the fundraising story here is also plays a part where they Corgi raised like a hundred and eight million series A and then a hundred and sixty series B. And then I love these like extra rounds, a series B one, another hundred million. So like, that's, that's a lot of cash. There's like three hundred and sixty odd mil, it's three. It's a lot of tokens. Yeah, it's a lot of tokens, three hundred and seventy odd million. And then they come out with this sort of vibe coded clone product. You still have to ask like, is the fundraising environment like hotter than the product market right now? I would, I would argue it is. And the pressure, I guess, is like, you know, hey, I have not gone through Y C, but hey, imagine I've gone through Y C and I'm, you know, thrown up to three hundred million.

• 23:29Speaker 00

  Of course, I, I feel the pressure to like, get a product out that's like, why shouldn't it be the same as, like the same or better than my competitor? Like, it's just, this is just a very strangely, quote, lazy way to do it, I think. Yeah. I mean, I think it's a confusing time, I think, for founders, you know, building companies. It's like, how do you think about? Unique, your unique value and how do you actually build a company in this market? If it's not building a model or something like that, like where do you really create value that isn't easy to copy? Yeah. And then just to sort of round this one out, I mean, this, this is in contrast to the last sort of case of this was Pair AI,

• 24:08Speaker 00

  which I think it was like almost two years ago at this point. But yeah, they basically just ripped an open source repo and then just like passed it off as their product. And, and also they got into even to murkier water because I think they started lying about it inside Y C. They sort of got asked like, hey, like just tell us what this is. And then they actually still went deeper on the lie before they sort of came clean. And then, and Gary Tan had like really pumped them up on, online for like a few weeks and then was like, oh, actually, sorry, that was, that was true. I've got to apologize and, and say like, we've disconnected from Pair AI. So really messy, that one.

• 24:47Speaker 00

  It's probably hard as an investor too, right now, to, to kind of like navigate such an influx of companies. Like, how do you kind of differentiate between what's real and what's not real? I guess there's always, that's always been something that investors have to navigate, but it's probably particularly true because I think some of the things that they might have used, especially if to like evaluate early stage teams are not necessarily the case now. Like you can create a really compelling early stage product rather quickly at this point. And you probably have to look, you know, a lot deeper than just what the product experiences and things like that to sort of evaluate the opportunity. Yeah. For one of the first times in history, I would probably not want to be a VC right now. I don't know. I just, yeah, having to pick through all these products and like try and figure out like,

• 25:31Speaker 00

  how is it actually made? And like, what's the real quote USP here? Like that isn't vibe coded by someone else or can be vibe coded by someone else tomorrow. Like very challenging. So one of our final headlines this week, really just about SpaceX, but you know, lots been talked about SpaceX elsewhere. More, we're kind of more interested in their acquisitions. So some may have seen already cursor, there are probably some. People's favorite coding tool was actually purchased by SpaceX. We're going to get onto that in the main topic, so we're not going to like touch on that one too deeply right now. But SpaceX also acquired a company called Mesh.

• 26:12Speaker 00

  So this is more of a sort of hardware tech company that builds optical links like on the ground as opposed to. So this was ex SpaceX engineers who built optical links that connected Starlink satellites. But they sort of, they think that they've seen that the same technology can speed up data centers on the ground. So yeah, we're like with this sort of influx of cash from their IPOs, I mean, two big acquisitions. I mean, if I just look at the cursor amount of, I think it was 60, 60 billion. Yeah, I'll just call it. I think it's way overvalued, but like, that's just me.

• 26:46Speaker 00

  So I have to assume that maybe this was maybe also slightly overvalued in its purchase as well. So just kind of throwing, throwing money at it when they don't have to think too hard about money. I mean, that might be something, but I also think that if you kind of look at the potential of this. SpaceX is already selling compute to places like Anthropic, Google and so on. And then they can own like the actual network layer amongst this is kind of like where Nvidia is like very much owning sort of the GPU and the chips. If you can own the network layer, which is another huge area of, of cost for these data centers, then that could be a gigantic, like total addressable market. And if these data centers continue to grow and people are investing in them, like you're going to need ways of transferring the data.

• 27:31Speaker 00

  And you can own essentially that layer that's responsible for transferring data, that's, that's massive. Oh yeah. I mean, I don't, I don't sort of disagree, dispute why they would buy a company like this for sure. Yeah. I mean, actually cursor is more the one I'm like, we'll get onto sort of the, the whys there, right? But yeah, this one mesh certainly sounds like a quote sensible play from the sort of the why. I can't see if I can find the, how much they actually paid for mesh. Just come back on that one, but it still looks like a slightly more strategic purchase than, than cursor. We're going to get onto cursor and just dev tooling in general in the main topic. Second, our final news item, it did hit the the main news was just Anthropic launching another product which sort of starts to cut into specialized products.

• 28:15Speaker 00

  It's called Claude Science and this is, this is a product that enables rich scientific artifacts to be sort of fully reproduced and visualized. So, so Claude Science says like generates figures and manuscripts alongside the code that created them. It can natively render rich scientific artifacts, including 3D protein structures, genome browser tracks, chemical structures and more. And yeah, this is just, I guess we, we've seen this a few times now where Anthropic like cut into a specialized. Finance and legal. Exactly. We saw the finance. Exactly. Yeah. So I'm going to say I'm not super familiar with the sort of scientific software space, but I can imagine it's maybe a little bit stayed in places.

• 28:57Speaker 00

  And so something when you've got someone like Anthropic coming in and saying, hey, we can, we can just bring our like might into this area. Like that sounds maybe, maybe really helpful, but equally like if you're a specialized scientific software provider, this could look a little bit concerning as well. Yeah, potentially. I do think that at least some of these launches that Anthropic has done in the specific verticals, people have kind of over indexed on, oh my God, this is going to like change the entire industry. It really kind of reminds me of like the early days of Google when everybody was like, every time like Google moved to do anything with some random 20% project, everybody's like, oh my, well, I guess all those companies are dead.

• 29:38Speaker 00

  Like here comes Google, but I think part of it's like, you know, where are their focus areas and also like where are their expertise? Like clearly Claude code is so amazing because they own the model and then they also have a ton of people who are like actively using Claude code to both develop the Anthropic technology. And then also feed back into the software that they use to build the technology. So it's this huge massive feedback loop. I think there, I doubt that they have that when it comes to some of these other domains. And maybe they get there, but I'm a little bit skeptic. I haven't, I just haven't seen the. You hear the launch, people freak out about it. And I like, you don't hear anything about these particular tools after that.

• 30:18Speaker 00

  Yeah, yeah, that's a very good point. I mean, I haven't been tracking, say, the finance suite that Anthropic came out with. I think when you sort of move out, move maybe slightly outside of having been like a, I went back to builder mode, you know, before joint subbase. I was very alert to like what these companies were doing and how the, how the products were actually netting out. Cause it always affected you if you sort of felt like you're building something that Anthropic or OpenAI were going to come and sort of annihilate. But yeah, we'll sort of follow along. That might be another sort of deeper topic. Yeah. That we can do in the future, actually. So speaking of, of main topic, we've actually covered a lot of headlines today. So this might be a little bit shorter than usual, but we just want to look at the, if you want to call it the IDE wars, but round two.

• 31:03Speaker 00

  Cause last time we, we looked at this, it was, it was kind of interesting. Like we're only talking maybe like a year apart. But the IDE wars round one was like cursor versus windsurf. And in both cases there, it was like, well, they're both kind of derived off VS Code and then where do they go with this? And what's the tools that will sort of quote win the day? And like, we just are in such a different place now. Cursor has, as we mentioned, just been purchased by SpaceX. So I think really what we're focusing on here is just the fact that who actually owns your dev tool chain now? Like it's, it's this idea of, well, depending on kind of your, if you want to call like top of funnel for where you write your code.

• 31:49Speaker 00

  You're also buying into effectively like an ecosystem quite often run by that, that company. So the obvious cases are like Claude Code from Anthropic and like Codex from OpenAI. You know, you use those tools where you're basically buying into the models that they themselves. Run and sell you and and are making a lot of money from. And so you're kind of like, what is your incentive to break away? If you're if you're sort of seeing success, you know, from building with those tools. You're less likely to want to like break out of that because you're like, well, this is working so well. I don't know how well the other one will work and so on. And then Cursor, what's the strategic strategy there, I guess. Well, SpaceX has, you know, XAI its own.

• 32:32Speaker 00

  And Grok, no, sorry, Grok, there's two Groks. That's why I always get confused. There's one Grok who got bought by OpenAI, which was more infrastructure. And then there's the other Grok with a K, which is within SpaceX. That's more like the the chat, the chat version, I guess. But you're again, you're buying into that ecosystem now with, or it's assumed with Cursor that if you, if that's where you're going to be writing your code, you're probably going to be signing through to SpaceX models. Then there's the other option, which is you go fully open source and like open code. Has actually been like really ramping up. It's apparently one hundred and sixty thousand GitHub stars, seven point five million monthly active developers.

• 33:13Speaker 00

  And the point is, it offers, it's model agnostic. It actually, it gives you access to like seventy five providers. So you're not, in theory, you sort of quote locked in if you're using that sort of tool chain. So yeah, we're just going to back to this situation where you're sort of almost working in like a closed IDE environment versus open. I thought we'd like left that that town a while ago, but here we are. We're back again. Yeah, I mean, there was a time like now, it's like what you were saying in terms of with certain choices, you're buying into the whole ecosystem and kind of being vendor locked in. It's like, okay, I'm going to use Claude code, which means I'm going to use Claude as my model as well. And then I'm going to essentially have my contextual data associated with that particular development experience.

• 33:57Speaker 00

  And then it was similar at one point in terms of like languages and IDEs choices as well. It's like, I'm going to use Microsoft's version of C++ which means I have Visual Studio and I kind of locked in that ecosystem. Or, you know, even a bigger example would be something like the early days of C# like the only idea. IDE, I could write C sharp in and the only compiler available was also from Microsoft. And it happens to be code that is most easily run on a Microsoft Windows machine. So it kind of like you're buying into this whole ecosystem. And then over time, that that has changed where most of the languages, I basically have like every language became open source. A lot of the IDEs, if they're not free and open source, there is a free and open source version of that.

• 34:43Speaker 00

  And it became no one that I think feels particularly vendor locked in. I wonder if this is like sort of a common trend and wave that you have with any software innovation is that at the beginning, you have these kind of like locked in experiences because there maybe there's like, there's an advantage really to being able to build those. And it's harder to build sort of the separated decoupled version of that, but that happens over time. You, you mentioned the open source tools. Well, Open code, yeah. Open code, yeah. So there, I feel like every four to six months, there's like a new hot open source coding project that like gets all kinds of GitHub stars and grows really quickly. And then there's like a new hotness, you know, I would assume that eventually some of that stabilizes as well.

• 35:28Speaker 00

  Yeah, I think it probably will. I think developers are still open to experimenting with these like different tool chains. And as we covered, I think on last SCD news, just the cost. So, okay, it's making lots of money for Anthropic and I think somewhat OpenAI, but certainly Anthropic are just like killing it right now on, on Claude Code and the amount of money that companies are spending through that. But I think, you know, what we were saying last month was, you know, what is the, what is the tipping point for companies actually saying, okay, like we've got to rein in spending here or you should go and find an alternative tool, even if it's a bit slower or like quote, the quality is a bit less, but like we can't sort of continue on this sort of

• 36:10Speaker 00

  just always going to be using the hottest, latest models. That's quite like also the most expensive. So the question. In with open code and kind of the model choices is there was like some head to head comparison on builder.io and like open code was apparently seventy eight percent slower than than Claude code. So I think we've got to kind of look at that. That that sounds pretty drastic, right? But like, I think if we Claude code is not exactly fast. Yeah. So I can see where developers just like if they're sitting waiting for this thing like open code to like chunk through with a with a model of choice. And they're just like, why am I doing this? Why am I not just using Claude code? Is it just because my employer doesn't want to pay for it?

• 36:52Speaker 00

  Yeah, I would say like, you know, maybe seventy eight percent slower is not ideal, but like some of the efficiency or latency issues is a little bit hidden because these experiences are async. Like a lot of people that I know that are fairly advanced, like at using the agent engine tools, you know, they're running four to six agents simultaneously. So they're giving direction and then skipping over to some other tab or experience and giving direction there. And it's a lot of like sort of babysitting these different agents. So it's not like you have to just say like, give a prompt and then go for a walk for thirty minutes, like the old days of, you know, compiling large pieces of source code. And then come back and it's like, oh, okay. Well, now now I put in my next response.

• 37:36Speaker 00

  Like, I think you can actually, you know, do something. And even if it churns away, it takes some time to to compute. You're doing other things in the meantime. Like, there's other things that you can be doing. It doesn't require a hundred percent of your focus. I think that is a big change, also in terms of the developer experiences, like classically, it's all about getting in like the zone, not being distracted. And now it's a lot of like context switching and like flipping between tabs in your in your terminal or different experiences within the within your IDE or something like that. So I think that changes things as well, but not to say that you you should probably aim to be a little bit faster. But there there's less concern there. But I think like, one of the things that's interesting about all this, though, is that so much of the value that you're creating

• 38:21Speaker 00

  when you're interacting with these tools is sort of the data and the context with which you're building around it. So. That becomes somewhat like the vendor lock-in even more than the model and the experience itself is like, if I needed this, like, what is the switching cost? Like, if I want to go from Quad Code to Anti-Gravity or Open Code, do I suddenly lose all this rich history and context that I've been building up? Like, this is my, this has become my buddy. Like, how do I manage? Like, and that's something that has always been an issue in all software. It's like, okay, that's why databases are so sticky, because the migration cost of a database is really expensive. But it's something that all companies ideally want to, I think, have the flexibility between vendors.

• 39:04Speaker 00

  And I wonder if there's going to be a market for creating like an abstraction layer, essentially, that manages all this context and data that gives you like a vendor-agnostic way of delivering that context across these different surface areas. Yeah, absolutely. I mean, yeah, I mean, I use Claude personally and we have a work subscription. But I've obviously got less, I've got more choice on the personal side in that sense. But I still, you know, the fact that it has memory and, yeah, I do kind of like, I'm just going to keep using Claude. Because, yeah, well, hey, it's got all my chats and it can reference things back and so on. And that definitely comes through on the code side as well. I think the question also here is, is just the neutrality piece with Cursor.

• 39:49Speaker 00

  You know, they were always seen as this like neutral company, quite frankly, because they were like VS Code, but we're a separate company. We've done a whole bunch of changes to the underlying VS Code open source. And then you can choose your models that you work from. And I think the big question now is, is like, A, will that continue? Or, you know, is Cursor Auto Mode, is that now always going to go through Grok? And is it always going to be on SpaceX-owned infrastructure? I can also just correct myself that, because Grok with a K is SpaceX. Grok with a Q, which was the other one I was referring to, it was actually Nvidia. They ended up partnering with Nvidia on the infra side, so not OpenAI.

• 40:31Speaker 00

  But yeah, Grok with a K, which is within SpaceX, yeah, it just means that if you're now using Cursor, what does that look like? And quite frankly, I used to use Cursor less so over time for various reasons. This is just like another reason I think I would start to think twice about using Cursor, unfortunately. Yeah. Clearly, there's also a market for companies that don't want to send anything potentially over public internet or to Cloud vendors as well, where they're gonna be in an air gapped environment, or they run on prem banks and so forth. Like, I talked to lots of customers in our world that they're running their own L L Ms. And then they need

• 41:12Speaker 00

  a tool that's gonna like a engineering tool that's gonna work in that environment as well. So there's clearly a market there. So it doesn't necessarily need to be like one winner takes all. Yeah. And again, I think we should, we should do like a deeper topic on this on S E D news, but yeah, open weight models, like they are just becoming like dramatically cheaper. I mean, obviously you can run them in different places, but you know, roughly speaking, saying that like DeepSeek V four Pro is like come down to like, is it forty four cents per billion tokens? And whereas like Claude open, sorry Claude Opus four point seven is like for the same thing, like five dollars. So it's like, that's like, that's more like an A X

• 41:55Speaker 00

  difference to using Claude versus DeepSeek. And again, anecdotally, I think developers are saying like, well, we're actually starting to see some really good results coming out of these open weight models. So it's, it's definitely something we have to, to watch very closely because like, if, if it's a tipping point where whoever's like running these for a developer, that, that tool chain becomes as say, quote, as good as like Claude code. And you can, you know, convince your, say, company like, there's no challenges here. Like, it's just eight times cheaper. That's, that's pretty huge. Hmm. So yeah, I mean, I think it's interesting to, to have looked at this in terms of, yeah, we're curious, like, what are you guys using out there? Audience, like always, like, if you've got very strong opinions on this, that we, we always love to hear them. So

• 42:40Speaker 00

  we're always just trying to sort of report back what we're seeing day to day. And, and I'd say to go from a year, if, if you'd said a year ago, like, basically Cursor and Win Surfer would almost be sort of quote, taken off the market. I would, I wouldn't have believed you, but so yeah, I definitely didn't predict Claude code and Codex, but Claude code especially, that's sort of taking over the I D space and, and that sort of C L I first. For all builders, you know, that really just seems to be winning, winning at the moment. Very interesting to see. But yeah, let's get on to our favorite part of the show, Hacker News Highlights. Do you want to kick us off, Sean? Yeah, sure. So I found this article that was about HackerRank had open sourced their ATS or application tracking system. Oh, right.

• 43:24Speaker 00

  And this guy did sort of a deep dive, Sam Bell, I think is his name. You deep dive into using the applicant tracking system to analyze its LLM capabilities of analyzing his own resume. So he had scored like 98 of 100, but he basically ends up stress testing the resume screener, running the exact same test over and over again, like 100 times. And he had scores that ranged from like 66 to 99. And then the articles were interesting because he brought, he like breaks it down, like where was the LLM sort of consistent? And where was it all over the place? So things where the LLM did a good job was, you know, a checklist of like, do you know Python?

• 44:06Speaker 00

  That's pretty easy or categories requiring judgment, though, something like, are these projects architecturally complex? Which a company might care about. They're completely subjective and basically the equivalent of a coin flip with the LM. So the takeaway here, of course, is LMs are great at parsing, they're great at matching, they're terrible at subjective evaluation, which you shouldn't essentially ask them to do, especially if you're depending on getting some sort of score associated with it. It's just not going to be dependable metric. Yeah, super interesting. Yeah, I wonder what would happen if I put my resume through that.

• 44:43Speaker 00

  Yeah, well, it depends. You might get anywhere from 66 to 99. Yeah, exactly. So on my side, we have, yeah, it's a partial port of Kubernetes to the browser, which is kind of interesting. So this is actually by, and it was another Sam, actually, but Engrock. So Engrock have kind of released this and what they stress is this is not like full somehow Gzip Kubernetes running in the browser. But it is a partial port of Kubernetes. Binary, so it's enough to run pods and and problem. Basically, it's got ports of several Kubernetes controllers. So it's got a pod scheduler, it's got namespace controller, Kube Proxy, deployment

• 45:24Speaker 00

  controller and a bit more. And then it's got a browser based take on, you know, this, the CNI, the container network interface. So like pods can talk to each other. And then it's got like a browser based container runtime, which the Kublet can talk to over the container runtime interface. So it's kind of, and yeah, I'm sorry, and just around an API for interacting with the webernetis, which is what they're calling it, webernetis cluster, to do things like apply manifests and and watch resources. So super interesting. I think this sort of angle here is is a little bit more on like education, for example, like it's quite difficult to learn Kubernetes. And so like, if you can do a lot of understanding how things operate and interact with each other in this sort of lighter, more interactive environment, I think that's, that's

• 46:09Speaker 00

  part of the the thinking here. But yeah, super cool. And the fact that it just comes via N-Grok, you can kind of assume it's actually pretty well put together. Yeah, that's fun. I also liked, I just have to highlight one comment that came through on that, which was, I think the name is Duncan GH, and they commented, investing early in this HN post before it's a banger, instant classic. So I just love that. Yeah, this is like, as this person says, a classic HN article. It's definitely why it made made my my highlights this week. Yeah, Hacker News was of like a port of a difficult technology to some other like four factor, just like, like the mirage of Doom ports that we've covered. Yeah, exactly.

• 46:53Speaker 00

  Yeah. And the last thing I had was this article, token maxing is dead. Long live token maxing. I saw that. Yeah, yeah. Yeah. So it's interesting. Basically, what the writer is saying is that originally, people were token maxing because we were encouraging developers to consume huge amounts of tokens. We're creating KPIs within companies for them. So a lot of it was like wasteful use of tokens. Essentially, what the author is saying is that now we're getting, we're maxing out tokens, not because of wasteful token maxing, but because actually something they term compound correctness. So as these coding agents improve, we actually spend more tokens increasing the likelihood of better outcomes. So if you spend sort of more time on compute,

• 47:36Speaker 00

  potentially even spend more money on expensive models, then it might cost you more over, say, a five-minute period versus a cheaper model. But with a cheaper model, you have to do more, you're basically wasting more tokens because you're doing more throwaway work. You gotta do more iterations, whereas you're able to get it right sort of the first time is the argument. Yeah, so the whole like one-shot still sort of holding up is more of like that's where to aim versus just rinsing tokens on cheap models, basically. Yeah. Yeah, my final one. It's a lot of Sams today. Am I, am I reading this correctly? Yeah, literally. So the NGROG one I talked about was, was by a guy Sam Rose. Uh, this is now Sam Wilkinson.

• 48:18Speaker 00

  Yeah, okay, so three Sams. Great. So this was RF hacking my cloud-controlled ceiling fan. So while like there was, it's quite, quite a practical application. This isn't just a sort of like off the deep end. This was like someone who had moved into, I think, a new house and there was a rickety old ceiling fan on there and they were like, right, I'm just gonna put a new one on. But as with most sort of smart devices these days, like it has to go through like a cloud service to run, like if you want to use any of the smart home features. So yeah, but I mean, I won't go into all the details, partly because I'm definitely no

• 48:58Speaker 00

  RF hacking expert, but you can be sure that this article has an amazing array of like how, you know, they decoded the RF signals and like figured out how to then set up a transmitter box that could then control this specific fan. And, you know, I definitely can relate to this. I have various ceiling fans where I live in Singapore. And if I could get them hooked up without a cloud connection, that would be kind of nice. So unfortunately, I don't plan to go through all the steps that. Sam Okinson did, but yeah, if you are so way inclined, then that's yeah, Sam Okinson to I O. So yeah, thanks for for posting that. Yeah, very cool. I wonder if like with all these tools that we're talking about, you know, cursor, Claude code, everything.

• 49:40Speaker 00

  If there be, it kind of opens up access for a lot of these like side hacker projects that maybe you could have done, but would take too much time and energy to do previously. Like, I haven't tried this. Like, can I have Claude like hack my Google Nest Cam to do something, you know, new or some, some piece of hardware or something that I wouldn't, you know, normally spend time on? Like, can I just send off an agent to do some of that work? Yeah, no, for sure. Maybe, maybe I'll have time. Looking ahead, I'm, I'm off to Scotland in a couple of weeks for, for the month, for the summer. So that will maybe give me some time to hack away on some things that I always struggled to. There you go. I expect me to be covering you in the Acronews session with your, your latest ceiling fan hack.

• 50:23Speaker 00

  Maybe that's a goal. Yeah. By the end of the year, something that one of us produces is, uh, in our side project world ends up there. But yeah, any other, I don't know, predictions, thoughts for the next month? I think we're gonna, this is not the last that we've heard about, like the government, you know, essentially non-US governments concerned about the US government controlling deployment of models. It would be my prediction. I think this is going to be a big topic of conversation for the, the foreseeable future. Yeah, for sure. On my side, yeah, I'm curious to see if like cursors, it's reported like cursors usage goes off, goes off a cliff. Maybe not. I'm, I'm just going to kind of put it out there. That's something I'll watch and, and we'll come back to.

• 51:06Speaker 00

  I'm just curious how a deal for this, A, was it really worth 60 billion? Sorry, I don't believe it was. And B, like, I think as a people who would try to avoid any product that is Musk related, just. To be honest, I do have a lot of like macro respect for him in terms of what he's achieved in many areas. But I think there's all sorts of other areas. I just couldn't agree less. And I think I've got so many other options on this product, I'm going to use something else. Yeah, yeah, makes sense. Right, well, a lot to cover there today. I'll be seeing you in Singapore next week, Sean. And we'll catch up then again on SED News in a month's time. So thanks. Thanks, everyone for tuning in. Thanks, everyone. Cheers.