Yes. Okay. So in all honesty, it was a bit of a silly LinkedIn-y type article. A lot of the finance world talk about the bull case and the bear case, the bull being the positive outcome, the bear being the negative. And I just thought it was quite a catchy title. But really, what the article was showing is we are in this weird Schrodinger's design career where we haven't opened the box yet. We don't know whether the design industry is alive or dead yet. And hopefully, you're right. In the next year or two, a new shape will form. Actually, before I get onto the bull and bear case, you know, because you mentioned something there, I wrote an article around the same time called The Six Levels of Adaptive Software, which I actually think really defines where we're heading. There are six levels of driverless car. You know, zero is basically you drive it yourself, five is that it automatically drives everywhere. I think people say that Waymo is a four and Tesla thinks it's a four or five, but really is probably a two. I think that is ultimately where we are heading with a lot, but not all of design. I think at the moment, we are sort of at level one where design teams are starting to explore around the edges and automate certain parts of the process. I think very quickly we will start to have systems that run human out-of-the-loop multivariate tests in the background. At the moment, if you want to do a multivariate test, a product manager or designer will come up with a test, they'll come up with the screens and you will run it through your multivariate testing system. Because most systems have quite a you know, most companies have quite a sophisticated design system. You don't really need a designer to actually assemble the parts of the multivariate test. And actually, if you've got a smart AI that's looking at all of your analytic, there's a really good chance that the AI can come up with tests. And actually, I'm seeing startups are doing just that. In fact, I'm seeing startups that know their system could run completely autonomously. But at the moment, product managers don't feel comfortable trusting it. And so they're basically kind of crippling it. They're almost saying, hey, look, you know, we have self-driving at the moment, but people don't feel comfortable self-driving. And so we're going to just do like lane assist or speed, you know, whatever. I think the first, the first or second level, I can't remember which the levels were, but very soon we're going to have automated multivariate testing. Once you've got automated multivariate testing, then we can say, well, look, multivariate testing is usually done for one user group. Why don't we self-optimize for five or six user groups? Rather than having a single local Maxima, why don't we have five or six local maxima? And so you start pushing out different designs for different user groups. Eventually, everybody will be a user group of one because, you know, the same way as Netflix, back in the old days, if I wanted to program a day's worth of TV, a human would have to program that and decide what came on where. Now I go to Netflix and the Netflix algorithm knows me, knows what I like. Generally, it gives me a really bad set of suggestions because even after 20 years, you know, Netflix and Amazon are still pretty bad at this, but it's better than what terrestrial TV does. I think we're moving to that way with design and AI, where I will come to a system, come to a shopping experience, let's say, and based on what the website knows about me, what the app knows about me, will assemble on the fly an interface pulled from all the component parts it has. It will know. In the UK, it will know what credit card I have. It will optimize the checkout flow. It will know the clothes I like if it's a clothing site. It might even have an image of me. And so it might pop me in a 3D avatar of the clothes so I see what it looks like myself. So, in this world, we're moving to this world where I think these systems will just design themselves. And so, I really think in five to 10 years, maybe 15 years, a lot of the big sites will be self-automating in the same ways in the manufacturing world. We talk about dark warehouses or dark manufacturing plants where there are so many automated robots and so few humans that you can turn the lights off because the robots just work without any humans there. We are moving to a world of dark product teams where the product team is an AI that does all this for you. So, this is where I think we're heading. I think that will mean that a lot of people in large companies won't be needed to do the design product management role. I think there will always be smaller companies that need that. You know, if you are using a Shopify account, then it will probably do that for you, but there'll be these smaller midterm brands. And so, that's the future. But you're right, that my bull and bear case is really the now to two years. We are in this really weird moment, which we haven't quite figured out yet. Whereby, if you are a product manager, you can now design. If you're a designer, you can now ship code. If you're a coder, you can now ideate and do kind of customer discovery and synthetic users to figure out what you want to build. And so, every single one of these distinct roles in the trifecta is now encroaching and creeping in to everybody else's role. Now, this is really interesting because I think for the designer, up until recently, a lot of designers were incredibly frustrated because we would see the things that were broken on our products. But in order to enact those changes, we had to petition. We had to petition for the product manager to put it on their carefully guarded backlog. We had to petition for engineering to get some time in their budget. We had to petition for all this stuff. And because design typically is lower in the hierarchy and also lower down the discovery chain of product, it was really hard for designers to get any time. And if we were ever given a little bit of time, it's up. Let's give the designers a sprint so they can do a bit of design debt management. Yeah, exactly. So, I think a lot of the time, or at least the last five or 10 years, design has been relegated to being Figma operators. Somebody further up the chain gives us a PRD. The PRD, you know, our product leaders think it's 90% of the solution. We get it and we realize it's 20% of the solution. So, already there's like a gulf there, but we basically dutifully go and try and fill out the PRD in Figma. And, you know, we'll be given a happy path. And we design a few non-happy paths, the exceptions, you know, the stuff that hasn't been filled in, and then we ship it and we do it again. I think AI changes that because now, in a vaguely communist Marxist way, we have been given the means of production. No longer do us designers have to go and ask capping hand for the boss, can we do this? Now, if we see some design debt and we have the tools and the capabilities to ship it because we're using AI, we can just see that design debt, we can fix a problem, we can put it into production, maybe we can launch it live, or at the very least, we put it into some holding pen where the engineers just look at it and go, you know, this is good enough. Similarly, like if we have an idea for a solution, rather than what we used to do is go to a meeting and give a really big like pitch deck or go and try and get six months' worth of research budget to prove our silly little idea was correct. Now we can go with a prototype that we have created over the weekend. Obviously, so has the product leader and so has the CEO and so has the CFO and so is the CFO's partner and CFO partners dog groomer. Everyone's coming. It's almost concerned of meetings now. Everyone turns up with a prototype and then it's which is the prototype that's better. But at the least we can see it. We're not having to go to them and say, hey, you know, can we get permission to do a week-long design sprint? Or can we get permission to get three engineers to build a prototype? And so we have much more control of showing viscerally our ideas. And I say, sometimes we might have to still get permission, but increasingly we'll just ship them. And so that for me was the bull case. The bull case is that designers are now potentially that the yoke has been taken off. The manacles have been taken off and we can ship. The single truth isn't a slide deck, a PowerPoint deck anymore. The single truth is what's live on the website. It's that whole ask permission rather forgiveness. It's much easier to go to your meeting and say, hey, look, I launched a thing and it's doing really well, rather than going, please, sir, may I have two months worth of resource on your already overwhelmed backlog. And so this is for me the bull case. The bull case is the brakes are off, the handcuffs are off and everyone is racing to ship. And I think that's really, really exciting. The risk is that we're slightly late to the party, you know, for really good reasons. The models really changed the end of last year, and the engineers were the first people to start using and start shipping them. And then, after the engineers, the product leaders got on board because a lot of the product leadership stuff is more thought partnery, it doesn't require dedicated tools. And it's taken a while for the pencils and papers of the world to catch up. And it's taken even longer, arguably, for the figmas of the world to make something that's vaguely half-decent. And so, I think a lot of the designs I meet are now nine months behind and now starting to adopt tools. They're either adopting the engineering tech stack or code or whatever, in which case they are basically becoming design engineers, and that requires a certain set shift. Or if you're a classic UX designer, you have been wondering to try and find the perfect tool set and you're slowly building out your design tool stack and you are now getting into being more productive. And so, um, we are a little bit behind. And one of the reasons why you I'm running this conference in the UK, Design plus AI, in September, because I want designers to realize where they are, realize they're a little bit behind, get up to speed really quickly. Because I do think there is a little bit of an arms race. I do think people are staking out their territory. And I think if designers don't do it, as has been shown over the years, other business partners will take that away. I mean, you know, you know, most of the designers I know are really nice, ethical, liberal-minded people who have big problems with the way that the training data was captured. They've got big problems with the ethical and moral way that the LLMs have been created. But if we sit around and debate that, someone else is going to eat our lunch. And so, I think we need to accept that the horse has bolted, that the data has been stolen, it's been trained, and it's there. The irony is obviously now all the big LLM companies in America are crying foul at the Chinese open source models, the open weight models, because they're doing exactly the same, which I think is a lovely little bit of Schauldenfreude. But I think we need to, irrespective of how we feel about the technology, I do think it's here to stay in some shape and form, and designers need to come in. And so, that's the bull case. And the bear case is opposite. The bear case is basically that everyone else, product management, you know, engineering takes over, they start doing all of the really interesting strategic stuff that we always want to do, but never can do. And ultimately, design sort of shrivels on the vine. And the main role of design in the future could potentially be the managers of the design system. You know, you might have like you might have a 200, 300-person engineering team, and you might have a five-person design system team that basically manages the design system that all of the engineers and product managers ship. And so, I think that is definitely the bare case that design shrinks down into being little more than the people that look after the design system.