Linear says AI isn't saving product teams time

Mike Belsito hosts Now Shipping, the weekly AI news show for product people, brought to you by the team at Mind the Product. This week's three stories cover a platform war breaking out over where the world's code lives, the first real fallout from AI content watermarking, and a set of behavioural numbers from Linear that quietly undercut the standard business case for AI adoption.

In this episode, we cover:
— Cursor launched Origin, a full GitHub competitor, on the same day GitHub suffered a six-hour worldwide outage with a 20% global failure rate.
—  Origin is deliberately interoperable with GitHub, which removes the switching cost that usually keeps frustrated teams on a platform.
—  Developer tooling is consolidating into a two-platform fight between Microsoft and SpaceX AI, which closed its $60bn acquisition of Cursor four days before the Origin launch.
—  Anthropic is now watermarking every response Claude generates, everywhere, in response to the EU AI Act's transparency code of practice.
—  The watermark uses SynthID Text, encoding a signature in low-stakes stylistic choices that survives light editing but not a full rewrite.
—  The backlash is loud but misdirected: other major model providers have signed the same code, so this is an infrastructure shift rather than a Claude-specific one.
—  Linear's first data report, drawn from over 127,000 paid users, shows AI adoption more than doubling in every function between January and June, with product managers climbing fastest.
—  AI now authors close to half of all issues created in Linear, up from fewer than one in a thousand two years ago.
—  Teams with a coding agent went from 21 pull requests a week to 65; teams without one moved from 8 to 10.
—  Planning time did not move at all, in any function, in a year when execution metrics rose everywhere else.
— Linear frames the result as a Jevons paradox: AI work landed as a new layer on top of existing work, and total time spent went up rather than down.

Chapters:

(00:00) Introduction 
(00:16) This week's stories 
(01:01) Cursor launches Origin 
(02:01) SpaceX AI's acquisition of Cursor 
(02:34) Interoperability and switching costs 
(03:04) A two-platform developer tools market 
(04:42) Anthropic starts watermarking Claude 
(05:15) What the EU AI Act requires 
(05:42) How the watermark works 
(06:35) The backlash 
(07:08) What it means for product builders 
(09:22) Linear's first data report 
(10:12) AI adoption across functions 
(10:37) AI-authored issues and pull requests 
(11:21) Planning time hasn't moved 
(12:30) Jevons paradox and rising workload 
(13:28) Wrap-up

Referenced:
Cursor: https://cursor.com
GitHub: https://github.com
Anthropic: https://www.anthropic.com
SynthID Text (Google DeepMind): https://deepmind.google/technologies/synthid/
EU AI Act: https://artificialintelligenceact.eu
Linear: https://linear.app
Linear's first data report
Mind the Product: https://www.mindtheproduct.com
Vote for Mike's SXSW session: https://tinyurl.com/belcetosxssw

Transcript

SPEAKER_00 0:00

Mike Belcito, and this is Now Shipping, the weekly AI news show for product people. Every week I bring you three AI news stories that matter to you as somebody who's actually building products for a living. Brought to you by the team at Mind the Product. This is Now Shipping. A preview of this week's stories on Now Shipping. First up, Cursor launched a direct competitor to GitHub on the same day that GitHub went down worldwide for six hours. As they say, timing is everything. Next, Linear published a data report this week based on real behavioral data from over 127,000 paid users, tracking exactly how AI is changing their product and engineering teams and how they work. But there's interesting things hiding in the data. And finally, Anthropic is now watermarking every response Claude generates. We talked about this last week, but now we're seeing the aftermath. All of that on this week's episode of Now Shipping. Now, for most developers, that question might not even feel real, right? Like GitHub has 180 million users. It's been around since 2007, and now it's owned by Microsoft. It's been billed as the place where the world's code lives. Replacing it almost seems unreal. And yet, this week, Cursor said, let's give it a shot. On August 18th, Cursor launched something called Origin, which is a brand new code hosting platform designed to do everything that GitHub does. Everything from repositories, pull requests, collaborative editing, forking, cloning, browsing. I mean, it does it all. It's very much a direct competitor to GitHub, and they launched it on the exact same day that GitHub suffered a six-hour worldwide outage with nearly a 20% failure rate across the globe. They couldn't have planned it any better. Now, four days before the Origin launch, on August 14th, SpaceX AI closed on its acquisition of Cursor in a $60 billion deal. So when Origin drops, it's not just Origin announcing new product, it's SpaceX AI basically making this big statement of where they want to be. Now, GitHub has had 257 documented outages in the past year. Right now, there's talk of this visible exodus of high-profile users leaving. Cursor is definitely having their moment right now. Now, an interesting wrinkle to this is that Origin isn't asking you to burn your GitHub down. It's actually designed to be interoperable. Your GitHub repos will sit alongside what you have in Origin. So it lowers the switching cost, which is usually the main reason that people stay on platforms that they're frustrated with. Cursor's saying you don't have to leave. Just try this alongside what you already have. And Cursor has promised agent native features for Origin that are coming, although, in fairness, they haven't exactly said what that really means yet. Now, there's a lot to unpack here, right? It seems like the developer tools market just became a two-platform fight between Microsoft and SpaceX AI. And think about what that means. I mean, GitHub is Microsoft, Cursor's SpaceX AI. It's two of the most powerful companies on the planet competing for where your team's code lives. And you could kind of see the two camps sort of gravitating towards the two platforms, almost like an Apple versus PC kind of thing. I think it's worth recognizing also that the tool your engineers use every day, the one where their custom workflows, their context, their muscle memory, all of that, is now owned by a company that's actively trying to do more on the developer stack. I mean, we just saw SpaceX AI launch Grockbot. Now it's into code hosting. When all of this starts to happen, it almost goes from like, oh, it's convenient to have all of these owned by one company to feeling like you're locked in. Now, again, on the timing, I think it's really interesting, right? I mean, this happened on a day where GitHub had this worldwide failure. Maybe it's luck. Or maybe SpaceX AI, they had it all ready to go and they were just waiting for their moment. They're just waiting to sort of pounce like a predator pounces on their prey. It would not put it past me to see Elon Musk leading SpaceX AI and waiting for that one moment where they could strike. And if that's the case, well, they played it well because they timed it all perfectly. Anyway, that is story number one. On to story number two. Alright, story number two is a continuation of a story that we reported on last week, and that is anthropic is now watermarking the text that Claude generates. Every response, it's invisibly embedded. So you may not be able to tell just by looking at the text, but people can figure it out. And now that there's been a few days to sort of let things simmer, there are definitely people out there that are furious about it. But let me give you a little bit of a refresher in case you missed last week's episode and you just missed the news in general. So as a result of the EU AI Act, Claude has begun watermarking text and other types of content that it generates, not just in Europe, but it's doing so everywhere. There is a part of the act called the Transparency Code, and it requires AI companies to make AI-generated content machine detectable. This is the law. And Anthropic, by the way, says other major model developers have signed the same code of practice. So it's not just a Claude thing, it's on its way to becoming an industry standard. And how that happens is when Claude is writing something, it sometimes faces low-stakes stylistic choices like whether to use the word overcast or gray to describe a cloudy sky. Now, those choices are essentially arbitrary from a quality standpoint. Anthropic system now uses those moments to encode a pattern, a specific invisible signature woven in the word choices that's undetectable to a human reader, at least so they say, but it's detectable to anybody who has the key. Now they're using an approach called synth ID text, which was developed by Google DeepMind in 2024, and they're releasing a detection API. Now, again, it's not meant to be so obvious, but it's not easy to remove. You know, if you tried to just do some light editing, that watermark stays intact. Only a complete rewrite will actually get rid of it. Now, when this news came out, all of the social platforms basically exploded. There are so many people that basically vowed to cancel their subscriptions. And there's a common theme. I mean, most people are upset that this will essentially out them in using AI, whether it's at work or at school. One user described it as putting a quote digital tattoo on their forehead for anyone that uses Claude to help with assignments or professional work. But then there was a counterpoint. Somebody said, Well, if the only reason you wouldn't want this is to lie to people, is this such a bad thing? Now, two sides to it, right? Now, why does it all matter? Well, I'd say it's a trust and transparency shift at the platform level. If you're building a product that uses AI to help users generate content, emails, reports, code summaries, in anything, your output will soon be detectable as AI generated, and your users might care a lot about that. It's also important to point out, again, it isn't just an anthropic thing. When the EU AI Act requires something of all major model providers, and all of those providers have signed the same code of practice, this is an infrastructure shift. So it's funny to see people like, oh, I'm gonna cancel my anthropic subscription. Well, it's really kind of for everybody. Now, I think the backlash is useful data, right? I mean, it's there are a meaningful number of AI users that are basically counting on nobody being able to tell that they were using AI generated content. Now, that group is now gonna have a harder time to do what they were doing before. Now, how you respond to that as a product builder, it all depends on what your product's helping them do. And look, my take is that it's not a bad thing to use AI for generating content. There is a thoughtful way to do it, there's a non-thoughtful way to do it, but you wanna be able to stand behind any content you're creating, whether you used AI to help generate it or not. The key here now is that there is going to need to be more transparency from everybody, from the AI models, from the companies and product people using those AI models to build products, and now the users that are using that AI generated content. So think about what your users are using your products for, what they expect from you when it comes to that kind of transparency, and what people might expect from your users when it comes from that transparency. And you can help potentially lead the charge with your users on saying, hey, transparency, it's not such a bad thing. There is an upside to all of this. Um, I think there are probably many more stories to come about this topic and anthropic, and again, all the other major models out there that are going to have to comply with this law as well. All right, that's story number two. One more story to go. All right, story number three is a little bit different than the other two stories we have for this week. It's not about a product launch or some sort of acquisition, it's about data, real data from a real product company about how AI is actually changing the way teams work. And it may even be the most impactful story we have this week because of what the data says. Now, you're probably familiar with the company Linear. That company has a tool that's a very popular project management tool with software teams. They've published what they're calling their first data report this week. Now they have over 127,000 paid users, and they've been watching exactly what happens inside the product as AI adoption has grown. And I'll walk you through some of the numbers because it's pretty interesting. First, between January and June of this year, AI adoption more than doubled in every function. Product managers went from 12% active on AI features to 34%. That's actually the fastest climbing function in the entire report. Go to market teams, the furthest from the code base went from 5% to 18%. And CEOs at companies of over 200 people went from 9% to 36%. That's the biggest jump in the whole data set. Now, here's something that's really interesting. AI is now authoring nearly half of all issues created in linear. Two years ago, it was fewer than one out of a thousand. Now it's approaching 50%. Now, pull requests, actual code shipped, that's up 111% over the past two years. Teams that connected a coding agent went from 21 pull requests per week to 65. But teams without that coding agent, they went from 8 to 10. So it went up, but most of the growth is happening on the agent side of things. Now, non-engineers are shipping code at a rate that would have been unthinkable two years ago. Product manager pull requests tripled from 3% to 10%. Now, buried in all of these numbers are two findings that I think people might skim past, but they might be the most important numbers yet. First, planning time didn't actually move. Time spent on customer requests, documentation, projects, that all held flat in every single function. In a year where execution metrics shot up everywhere else. AI is changing how fast teams build, but it's not changed how teams figure out what to build. The judgment work, the talking to customers, deciding what actually matters. It's taking as long as it always has. And this makes a lot of sense to me, actually. It shows that you can automate so much of the mechanics of product work away, but AI won't help your team decide what to build. That's actually a core thesis of the book I'm writing right now, The Future Proof Product Manager. It's due out from Wiley, February 2027. In fact, it's the book I'm trying to launch at South by Southwest next year. Link in the show notes. All right, my shameless plug is over. But again, that stat that actually makes a lot of sense when you think about it. Now, the other big finding that Linear had is that work has gone up. Linear explicitly calls this out as a Javon's paradox. Time spent on AI tasks, chatting with AI, running agents, it appeared as a new layer on top of existing work. Nothing shrank to make room for it. Overall, there wasn't time saved on product development. In fact, the time spent went up. So if you're building a business case for AI adoption around this, like, oh, we're gonna save people time, you're gonna do less work. The data says that's not what's happening, at least not at the teams that Linear is watching. What's happening is that teams are doing more with the same number of people, which might be the right outcome for your company, actually. Or it might mean that you're adding to your team's load and you're calling it efficiency, but it's not actually efficiency the way that you're trying to define it. So interesting stats coming out of that report. And that is gonna put a wrap on this week's episode of Now Shipping. If you found this valuable, please tell a friend, let somebody know about it, subscribe. I would love your support in continuing to get the word out about this show. And if you have suggestions for me, leave a comment below. I promise you I will read the comments that you leave. And I will see you here next week with three new AI stories that matter to you as a product person. Once again, my name is Mike Belcito, and brought to you by the team at Mind the Product, this is now shipping.