The Exchanges

Q Can I be direct? I mean, this is a nice way. I don't understand. How does building for agents change the way that we build?

A Let me take some examples for, for stuff. We've done that's been successful with agents, right? Like, so one thing that we built specifically to improve our agent experience was building an API flow that would allow an agent to create a site and deploy a site without any users signing up for Netlify first, and then hand over that site to a user later with a claim flow. And again, simply from the realization that when we have a normal developer, The first creation of the site will always be triggered by that developer doing something with Netlify, right? Like they'll pick a template and start, or they'll Link to a new GitHub repository or they'll deploy from their command line. But once you have a CodeGen agent, they might just ask that agent, build me a web app with a Sudoku game, whatever. And now it's shared with people on a URL. So we gave those agents the ability to just deploy to Netlify, no sign up from the user, show the user the URL and say like, here's what is built for you. And then later the user can go Claim it and say, okay, I actually want a Netlify account or I have a Netlify account added here. I'll, I'll want this to stick around. I'll put a custom domain on it and so on. Right. We've seen a lot of companies now start adopting similar flows, right? Like in the authentication space, in the database space and so on to make the tools available to agents without an…

AI assessment note: “building an API flow that would allow an agent to create a site”

Q Can I be direct? I mean, this is a nice way. I don't understand. How does building for agents change the way that we build?

A Let me take some examples for, for stuff. We've done that's been successful with agents, right? Like, so one thing that we built specifically to improve our agent experience was building an API flow that would allow an agent to create a site and deploy a site without any users signing up for Netlify first, and then hand over that site to a user later with a claim flow. And again, simply from the realization that when we have a normal developer, The first creation of the site will always be triggered by that developer doing something with Netlify, right? Like they'll pick a template and start, or they'll Link to a new GitHub repository or they'll deploy from their command line. But once you have a CodeGen agent, they might just ask that agent, build me a web app with a Sudoku game, whatever. And now it's shared with people on a URL. So we gave those agents the ability to just deploy to Netlify, no sign up from the user, show the user the URL and say like, here's what is built for you. And then later the user can go Claim it and say, okay, I actually want a Netlify account or I have a Netlify account added here. I'll, I'll want this to stick around. I'll put a custom domain on it and so on. Right. We've seen a lot of companies now start adopting similar flows, right? Like in the authentication space, in the database space and so on to make the tools available to agents without an…

AI assessment note: “building an API flow that would allow an agent to create a site”

Q I totally get you. When you look forward to that future, open versus closed is a very big question. How do you think about which approach dominates in an agent first world and why?

A Yeah. I mean, I've, I think every time there is a platform change, we see this battle between open versus closed. This is another kind of platform change. I think open will win again, in part because I think the ones that were best positioned initially to become fully vertical lockdown platforms was really like the large language model makers. And I think there was a moment where it really looked like we would just have very few of those that would be incredibly important and like would be able to sort of build a whole stack around them and that was it. And then suddenly model training costs kind of went down and like This open model came around and DeepSea came around and so on, right? Like now it seems a bit the opposite that like, there'll be a lot more to do to build agents that can then swap out whatever model is best and combine the models together and so on. And it'll actually be an advantage for like the agent builders to not be tied to a specific model and be able to like, just jump around between whichever model is best. And I think once that start is happening, we're going to have like that, that's going to mean a much broader competing ecosystem of agents. And that's going to make the benefits of an open model where people can bring their own agent win out over a closed model where everything is like fully verticalized.

AI assessment note: “I think open will win again, in part because”

Q When I interview people today, Matt, the common public company short is Salesforce for a number of different reasons, but do agents merely turn Salesforce into a data repository?

A That's definitely the big threat to Salesforce, right? Like to just be reduced to some kind of system of record. I'm sort of mixed on whether long-term that's a really, really bad thing for, for Salesforce or if it could also be a good thing for them, right? Like if you really start becoming like the main system of record that all of these tools rely on that, that's like, I don't think you should underestimate just how valuable a position that is because all those flows and apps and everything will, will run on top of it. And sure, it might change a lot of what Salesforce look like today and to what they look like in the future. But they definitely feel it very strongly as like the biggest threat to them to be reduced to that system of record and not the actual like workflow platform. And I think that's pretty likely what would happen, right? Like I think what we're seeing now is definitely the cost of building custom workflows, custom UIs, custom applications is going to dive down dramatically. And typically that will mean that demand for doing that will Go up radically and people will start building much more custom applications for everything, right? Like, and then the question is like, where does all that data live that those applications need to, to interact with? And where do the rules that govern that data live to make sure that it's also safe for these agents to, to, to…

AI assessment note: “I think that's pretty likely what would happen, right?”

Q When you look at the different stages. We mentioned that lovable and, uh, you know, being able to create very easily. Prototyping is now like step one. Have an idea? Do design tools like Figma die in the next wave of build?

A They're definitely going to have to really think a lot about this, this, this full stream of how prototyping again is often going to shift from the design tool into building something. I still think for a certain level Again, we talked earlier about like, does brand matter? Does anything like that matter, right? Like as long as that still matters, it will also matter having teams, I think that work with tooling that operates as the visual level and the interface design level. But if Figma doesn't manage to connect that to the full end to end chain of prototyping and building and so on, then this could be a pretty existential threat. To companies like that over time, or at least like a threat to their current revenue model, because there's no doubt that the way people start projects in general are really already starting to shift a lot, both from traditional templates and bootstraps and front end frameworks on the one end, on the other end from like design prototypes and visual design on the other end, right? Like to, to, to prompting as, as one of the key ways to just start a new project.

AI assessment note: “this could be a pretty existential threat. To companies like that over time”

Q a specific company needs for a specific workflow that they have. I agree with you that dev shops that build specialized custom solutions will be replaced by your lovables of the world. But you said generic SaaS tools. And we have companies like Klarna who are saying, oh, we're replacing all of our SaaS with AI. They're not, they're absolutely not. And they won't. Why do you think they will?

A Again, I think it will be more and more common to start doing that, right? Like, because if you think of generic SaaS tools, right? Like they are also solutions to a specific business problem that businesses have. They'll just say, okay, this problem looks very similar across like Tens of thousands of businesses. So we'll build like the generic version that, that works across all, all of those businesses, right? Like, so they don't each have to build their own. And that comes with a set of trade-offs, right? Like you get some benefits from getting a generic tool that encodes some best practices and so on, right? Like, but you typically also have to adapt how you work as a business to the tool you just bought, right? Like you, you typically sort of have to, with a lot of the SaaS tools that we, I bought over time of, of building up Netlify, it's been like Oh, we're kind of doing things this way now, but this tool kind of says we should do things this way. So, so we're going to like start adopting to the tool because we can't afford to just build our own stuff. And that's where I mean that more and more things that are right now in the balance where you would buy a generic tool that works for lots of companies for more and more of those problems, it'll start being a viable solution to say, let's just build our own like that. Like let's just solve our specific problem instead of L…

AI assessment note: “it'll start being a viable solution to say, let's just build our own”

Q Final one. And then we'll do a quick fire. What behavior or action in the wave of AI that we've seen has been most what to you?

A I mean, I think it was like probably the panic over. AI safety when ChatTPT came around, right? Like it was like, oh, they're immediately gonna kill us all and take control of nuclear missiles and like out-compete humans and so on where it was like, wait, when you actually go and look at these LLMs, they're very, very far from any kind of like internal agency outside of like still give me a text that says this and I'll try to predict it. Right. Like that was probably like the most surprising to me that like Extreme panic around that for, for, for a moment with, like, all kinds of attempts of really, really strict regulation way too early.

AI assessment note: “panic over. AI safety when ChatTPT came around... most surprising to me”

Q a specific company needs for a specific workflow that they have. I agree with you that dev shops that build specialized custom solutions will be replaced by your lovables of the world. But you said generic SaaS tools. And we have companies like Klarna who are saying, oh, we're replacing all of our SaaS with AI. They're not, they're absolutely not. And they won't. Why do you think they will?

A Again, I think it will be more and more common to start doing that, right? Like, because if you think of generic SaaS tools, right? Like they are also solutions to a specific business problem that businesses have. They'll just say, okay, this problem looks very similar across like Tens of thousands of businesses. So we'll build like the generic version that, that works across all, all of those businesses, right? Like, so they don't each have to build their own. And that comes with a set of trade-offs, right? Like you get some benefits from getting a generic tool that encodes some best practices and so on, right? Like, but you typically also have to adapt how you work as a business to the tool you just bought, right? Like you, you typically sort of have to, with a lot of the SaaS tools that we, I bought over time of, of building up Netlify, it's been like Oh, we're kind of doing things this way now, but this tool kind of says we should do things this way. So, so we're going to like start adopting to the tool because we can't afford to just build our own stuff. And that's where I mean that more and more things that are right now in the balance where you would buy a generic tool that works for lots of companies for more and more of those problems, it'll start being a viable solution to say, let's just build our own like that. Like let's just solve our specific problem instead of L…

AI assessment note: “it'll start being a viable solution to say, let's just build our own”

Q Do you think you mentioned prompting that? Do you think chat is the right interface for this next wave of AI? I've had many guests on the show before that say the worst thing with chat GPT is it made chat and prompting the default interface. Do you agree?

A Yes. I don't think chat is like the B and all of interfaces. It's a very sort of simple transactional flow that that's easy to, to work with. And I think chat or conversational interfaces will, will be a thing, right? Like it's also a thing. Between humans, right? Like we all use Slack and Messenger and WhatsApp and all of these chat tools for good reasons. It's not like that will go away, but I think there's a whole other set of user experiences and user interactions. That we have in sort of the non AI powered world that we haven't really yet figured out how, how to do in the AI powered world. So one, one of the things I've been talking a lot about is this path where right now, when we look at all this custom development and all of these apps like bold and lovable and so on, right? Like right now they're generally making it a lot faster. To build exactly the same things on the web that we built the last 10 years, right? Like the same kind of applications, the same kind of marketing sites, the same kind of user experiences that we built through the last 10 years, right? Like, and we just like driving down the price of building that. But the next frontier for me is sort of figuring out what is it we can build in terms of actual user experiences when we have LLMs and AI available that we couldn't build. Two to three years ago, because I think there's a, that there's a lot of, of …

AI assessment note: “I don't think chat is like the B and all of interfaces.”

Q When I interview people today, Matt, the common public company short is Salesforce for a number of different reasons, but do agents merely turn Salesforce into a data repository?

A That's definitely the big threat to Salesforce, right? Like to just be reduced to some kind of system of record. I'm sort of mixed on whether long-term that's a really, really bad thing for, for Salesforce or if it could also be a good thing for them, right? Like if you really start becoming like the main system of record that all of these tools rely on that, that's like, I don't think you should underestimate just how valuable a position that is because all those flows and apps and everything will, will run on top of it. And sure, it might change a lot of what Salesforce look like today and to what they look like in the future. But they definitely feel it very strongly as like the biggest threat to them to be reduced to that system of record and not the actual like workflow platform. And I think that's pretty likely what would happen, right? Like I think what we're seeing now is definitely the cost of building custom workflows, custom UIs, custom applications is going to dive down dramatically. And typically that will mean that demand for doing that will Go up radically and people will start building much more custom applications for everything, right? Like, and then the question is like, where does all that data live that those applications need to, to interact with? And where do the rules that govern that data live to make sure that it's also safe for these agents to, to, to…

AI assessment note: “That's definitely the big threat to Salesforce, right? Like to just be reduced to some kind of system of record”

Q I totally get you. When you look forward to that future, open versus closed is a very big question. How do you think about which approach dominates in an agent first world and why?

A Yeah. I mean, I've, I think every time there is a platform change, we see this battle between open versus closed. This is another kind of platform change. I think open will win again, in part because I think the ones that were best positioned initially to become fully vertical lockdown platforms was really like the large language model makers. And I think there was a moment where it really looked like we would just have very few of those that would be incredibly important and like would be able to sort of build a whole stack around them and that was it. And then suddenly model training costs kind of went down and like This open model came around and DeepSea came around and so on, right? Like now it seems a bit the opposite that like, there'll be a lot more to do to build agents that can then swap out whatever model is best and combine the models together and so on. And it'll actually be an advantage for like the agent builders to not be tied to a specific model and be able to like, just jump around between whichever model is best. And I think once that start is happening, we're going to have like that, that's going to mean a much broader competing ecosystem of agents. And that's going to make the benefits of an open model where people can bring their own agent win out over a closed model where everything is like fully verticalized.

AI assessment note: “I think open will win again, in part because”

Q I'm like, it doesn't even do a Twitter thread. It's not exactly going to do globalization. Exactly. I feel safe for the time being. Listen, Matt, I want to do a quickfire round, so I say a short statement, you give me your immediate thoughts. What do you believe that most around you disbelieve?

A One thing is probably the sense that we'll still actually be building UIs and applications and websites and so on, where I think there's a lot of people that think that once we have agents, everything will be disintermediated, then the agents will just do stuff with each other and so on. But I believe that as humans, as long as those agents are actually doing things because humans are trying to do things, humans need these like shared experiences that we can pass around linked to and that we can understand like, Hey, I'm seeing the same as you're seeing and so on. Right? Like, so that's probably one thing where, where sometimes See a lot of people thinking that all of that, it's going to go completely away. And I, I don't think so at all.

AI assessment note: “we'll still actually be building UIs and applications and websites”

Q Do you think you mentioned prompting that? Do you think chat is the right interface for this next wave of AI? I've had many guests on the show before that say the worst thing with chat GPT is it made chat and prompting the default interface. Do you agree?

A Yes. I don't think chat is like the B and all of interfaces. It's a very sort of simple transactional flow that that's easy to, to work with. And I think chat or conversational interfaces will, will be a thing, right? Like it's also a thing. Between humans, right? Like we all use Slack and Messenger and WhatsApp and all of these chat tools for good reasons. It's not like that will go away, but I think there's a whole other set of user experiences and user interactions. That we have in sort of the non AI powered world that we haven't really yet figured out how, how to do in the AI powered world. So one, one of the things I've been talking a lot about is this path where right now, when we look at all this custom development and all of these apps like bold and lovable and so on, right? Like right now they're generally making it a lot faster. To build exactly the same things on the web that we built the last 10 years, right? Like the same kind of applications, the same kind of marketing sites, the same kind of user experiences that we built through the last 10 years, right? Like, and we just like driving down the price of building that. But the next frontier for me is sort of figuring out what is it we can build in terms of actual user experiences when we have LLMs and AI available that we couldn't build. Two to three years ago, because I think there's a, that there's a lot of, of …

AI assessment note: “Yes. I don't think chat is like the B and all of interfaces.”

Q Final one for you. When you look forward, 10 years at the most exciting things that happen, What excites you most in the next 10 years of AI?

A The capabilities it can give for human creativity and for building like amazing shared experiences, right? Like what it can do for builders seems really amazing, right? Like I always collect examples of, of people that are starting to build things like as one person or something like that, that are just like, wow. Right? Like this illustrator that's making these like tiktok sci-fi epos with like ai generated video that's just like wow one person can just sit and come up with with something like that right like as i mentioned this ux designer from niantic that can build like a full three d editor in the browser just by now if you'd like to watch that episode you can of course find it on spotify by searching for 20 vc that's 20 vc on spotify Prompting, right? Like, and I keep thinking like, if we do things right and we make sure that open platforms keep being really accessible and readily available, we really invest in tooling for people to, to build stuff and own it themselves and be creative. AI just has, has so many exciting abilities to let humans be more creative and do more interesting things. That's what excites me the most, right? Like that's why I built tools for builders in the first place.

AI assessment note: “The capabilities it can give for human creativity and for building like amazing shared experiences”

Q When you look at the different stages. We mentioned that lovable and, uh, you know, being able to create very easily. Prototyping is now like step one. Have an idea? Do design tools like Figma die in the next wave of build?

A They're definitely going to have to really think a lot about this, this, this full stream of how prototyping again is often going to shift from the design tool into building something. I still think for a certain level Again, we talked earlier about like, does brand matter? Does anything like that matter, right? Like as long as that still matters, it will also matter having teams, I think that work with tooling that operates as the visual level and the interface design level. But if Figma doesn't manage to connect that to the full end to end chain of prototyping and building and so on, then this could be a pretty existential threat. To companies like that over time, or at least like a threat to their current revenue model, because there's no doubt that the way people start projects in general are really already starting to shift a lot, both from traditional templates and bootstraps and front end frameworks on the one end, on the other end from like design prototypes and visual design on the other end, right? Like to, to, to prompting as, as one of the key ways to just start a new project.

AI assessment note: “if Figma doesn't manage to connect that... then this could be a pretty existential threat”

Q You can invest in Anthropic at 60, OpenAI at 300, or Grok at 50. Which would you buy and which would you sell and why?

A I would probably go for Anthropic. It's a really interesting thing from the outside where a lot of their motivation started around these like safety concerns and like, how do we, how do we allow the models to not take over the world and so on. And I think after sort of an initial panic that like, now that seems pretty like, okay, like LLMs right now are probably not going to like escape and take over the world. I think it turned out that a lot of the mechanisms you needed to build to make sure that the model's output was safe. Applies also to making sure that the model's output is usable for a practical purpose. And I think that's where Anthropic have really built strength, right? Like in terms of actually being able to guide their models to do things that work in production or to write the right code or to Like not break the syntax and stuff like that. And, and I think if they can keep on that, it, it, it makes their models really real world applicable.

AI assessment note: “I would probably go for Anthropic.”

Q So would you sell open AI or Grok then?

A I would probably sell Grok. I still think like it sits a little more in a, in a, in a middle space where, as you said, open AI on the other thing has just like really captured. I think it is just when you as a consumer think about going and asking an AI for something, it is still. Just the brand you go to, right? Like it is the first one you think of and I think that carries very, very real value and it's very hard to build with the scale of users they have and with like, just like the ingrainedness of chat TPT into the language as like, what you would call, like what a normal person would talk about if they talked about going and talking to an AI, right?

AI assessment note: “I would probably sell Grok.”

Q Do you think we'll see a specialization of LLM? You see Anthropic really excel on developer tooling in particular and code, but then honestly, open AI just wiped the floor with them on consumer touch points, consumer brand.

A Yeah. I mean, I think, I think you just answered it in a way, right? Like I said, I think we're already starting to see models get a brand as being good at different things, right? Like, and you start building affinity to, okay, if I'm doing this thing, I'm probably going to go to this type of model and this company seems to like really Really served me there, right? Like, and it thinks again, it's so early that who knows, right? Like, but the sign right now is that yes, we, we will start seeing large language models sort of have, have different strengths and weaknesses and, and you will go to some of them for, for specific use cases and, and Tropic definitely has, has really stood out in, in their ability to, to solve programming and coding style tasks in a way that I think is like I'm very bullish on Anthropic for that reason, because I think that's such a groundbreaking part of what makes LLMs multiply what they can do by, by a lot, right? Like the better they are at programming other computers to do stuff, the further the reach of what they can actually do.

AI assessment note: “we will start seeing large language models sort of have, have different strengths and weaknesses”

Q And speaking of like carrying value and going full circle, we spoke a lot about the Asian experience. How does value accrue in an agent economy? If, like, an agent builder has to pay you and Bolt or Lovable on OpenAI, how much margin is left?

A Every layer will shift spend from somewhere else in a way, right? Like that whole layer of bold, unlovable, and also cursor and windsurf and all of these toolings, right? Like they will shift budget that would normally go into more hours of developer time to build anything. And there's a lot of that budget that can shift into building more things instead, right? Tools like ours will shift budget from Maintenance, operations, DevOps, like the actual cost of running a thing after you build it. There's a lot of budget there to shift into paying a platform that, that works. And then if that's true, you'll kind of still have the, the, the same amount of budget available for, for anyone buying your thing or getting value out of, out of your thing, right? Like, but, but the pieces of budget that will move is like the hours you would have to pay developers before to, to build something.

AI assessment note: “Every layer will shift spend from somewhere else in a way, right?”

Q Why? Do you not go into a shop and ask for what you want?

A Yeah, but you also have like, you think about how, how much a fancy brand spend on that shop, right? Like, and on that, this one, the display windows and, and on people that don't ask first, right? Like on everything that's supposed to draw you into the point where you're ready to ask like, Hey, can, can, can I, can I have a look at this, at this thing, right? Like that, that's a, that's a very expensive industry, right? Like to, to make all of that happen. And I think, the more the world Goes digital, right? Like if you want to build any kind of that kind of brand affinity and like storytelling, you also got to find ways to do that. Today, honestly, like most e-commerces, for example, like the last three years have kind of been very stagnating in terms of like development in that field, right? Like, and everything looks, looks pretty, pretty similar with, with, with some changes, right? Like, but I think when the cost of building an e-commerce that looks like any of the current brands today goes down to like, just Ask Bolt to do it for you. Ask Lovable to do it for you. And you have that. Those brands are going to have to come up with, with something much more engaging if, if they want to keep stand out and if they keep wanting to build relations to people as more than just like a one-click shopping experience amongst any other.

AI assessment note: “display windows and, and on people that don't ask first”

Q You can invest in Anthropic at 60, OpenAI at 300, or Grok at 50. Which would you buy and which would you sell and why?

A I would probably go for Anthropic. It's a really interesting thing from the outside where a lot of their motivation started around these like safety concerns and like, how do we, how do we allow the models to not take over the world and so on. And I think after sort of an initial panic that like, now that seems pretty like, okay, like LLMs right now are probably not going to like escape and take over the world. I think it turned out that a lot of the mechanisms you needed to build to make sure that the model's output was safe. Applies also to making sure that the model's output is usable for a practical purpose. And I think that's where Anthropic have really built strength, right? Like in terms of actually being able to guide their models to do things that work in production or to write the right code or to Like not break the syntax and stuff like that. And, and I think if they can keep on that, it, it, it makes their models really real world applicable.

AI assessment note: “I would probably go for Anthropic.”

Q So there are devs using cursor, there are devs using lovable and bolt, and then there are non devs using your lovables and your bolts. When you look across that stack, how many do you think remain in the devs using cursor and your alternatives?

A Because I've been in this space of building tools for web developers, like for, for way too long, I guess, right? Like I've, I've kind of. I've seen the same discussion forever, right? Like, because early on when I was CTO of a company in Spain, we were like mass producing websites for small to medium businesses in Europe and so on, right? Like, and I was a CTO building the whole platform that, that our freelance designers would use to do that work on and everything, right? Like, and tools like Squarespace, Weebly and so on started coming on and everybody was saying, oh, but like, that's going to be the end of all web development. People are just going to use those tools, right? That didn't happen, right? Like it just lowered the barrier to entry. It meant that every little corner store could have a website, but it also just pushed up like what, what you needed to show up with if you're somewhat more professional. And then I started this whole Jamstack movement around decoupling the actual web front end from, from the backend and starting to build it at that level, right? Like, and we saw probably in the years since like, 2016, 1718, we started seeing all these Bootcamps, just like starting to spit out, react developers. And there was like this sentiment that they're not real developers and they don't know the whole stack and so on. Right. Like, but it turned out that they were…

AI assessment note: “the layer below... is probably still at least as large as this was before”

Q I would love to stop. How does building for agents, not humans, change the way that we build?

A Yeah, so the first thing I would say is that when I think about building for agents, I still start by thinking about building for humans. In essence, you could say that there's two versions of that 10 year future. There's one where, where the agents just kind of escape that they're just off doing their own things and we're sort of left behind, right? Like I hope that's not what happens. But there's another one where agents are tools that all of us use to do stuff with, right? Like in where agents are fundamentally always Doing something because some human is trying to accomplish something through an agent. And now we are in the very early days of building a discipline around agent experience and really starting to think about like when a user tries to get an agent to do something through your product, what happens?

AI assessment note: “building a discipline around agent experience and really starting to think about”

Q Is that not just a Google product strategy? So I would say that Facebook is the same like Lama and the productization around Lama has been bluntly very, very poor at the moment.

A And I think most of the, and that's what I mean. I think most of the tools I use. That Identic sits outside of any of those walled gardens, right? Like, and we're starting to see things like a, you'll see that there's, for example, a Chrome extension to use ChatTBT within Google Docs instead of Gemini, just because people want it, right? Like, because like, Hey, I have, like I use ChatTBT all the time. It has my history. It knows me now and so on, right? Like I want to bring it there, right? Like you start seeing those things emerge as like freshers. To do stuff with, with these products, with other agents, because the vertical strategy, at least right now is producing, I think more planned and worse result than, than the myriad of companies just like in building, agentic experiences, right? Like, so right now my, my signal is that like the open world will, will probably create more compelling innovation that people will want to bring into the product they otherwise use. And if they really can't, they might start using other products.

AI assessment note: “the vertical strategy, at least right now is producing, I think more planned and worse result”

Q So there are devs using cursor, there are devs using lovable and bolt, and then there are non devs using your lovables and your bolts. When you look across that stack, how many do you think remain in the devs using cursor and your alternatives?

A Because I've been in this space of building tools for web developers, like for, for way too long, I guess, right? Like I've, I've kind of. I've seen the same discussion forever, right? Like, because early on when I was CTO of a company in Spain, we were like mass producing websites for small to medium businesses in Europe and so on, right? Like, and I was a CTO building the whole platform that, that our freelance designers would use to do that work on and everything, right? Like, and tools like Squarespace, Weebly and so on started coming on and everybody was saying, oh, but like, that's going to be the end of all web development. People are just going to use those tools, right? That didn't happen, right? Like it just lowered the barrier to entry. It meant that every little corner store could have a website, but it also just pushed up like what, what you needed to show up with if you're somewhat more professional. And then I started this whole Jamstack movement around decoupling the actual web front end from, from the backend and starting to build it at that level, right? Like, and we saw probably in the years since like, 2016, 1718, we started seeing all these Bootcamps, just like starting to spit out, react developers. And there was like this sentiment that they're not real developers and they don't know the whole stack and so on. Right. Like, but it turned out that they were…

AI assessment note: “I've seen the same discussion forever, right?”

Q I would love to stop. How does building for agents, not humans, change the way that we build?

A Yeah, so the first thing I would say is that when I think about building for agents, I still start by thinking about building for humans. In essence, you could say that there's two versions of that 10 year future. There's one where, where the agents just kind of escape that they're just off doing their own things and we're sort of left behind, right? Like I hope that's not what happens. But there's another one where agents are tools that all of us use to do stuff with, right? Like in where agents are fundamentally always Doing something because some human is trying to accomplish something through an agent. And now we are in the very early days of building a discipline around agent experience and really starting to think about like when a user tries to get an agent to do something through your product, what happens?

AI assessment note: “we are in the very early days of building a discipline around agent experience”