20Product: How to Design and Build Products in a World of Agents | Why AI Will Kill Many SaaS Products | What Products Will Thrive and Die in a World of 100M Developers with Matt Biilmann, Co-Founder and CEO @ Netlify

0:00I think we'll see the next 100 million software developers emerge around tools that are authentic, but I still think it will be a profession to use those tools to do stuff, to build stuff for other people. If I look at Google's products, like I now have all these Genesis buttons all over and I never use them, they just haven't managed to productize it in a way that feels super compelling. This is 20 product with me, Harry Stabbings. Now, stay we focus on a core new focus for product teams. How to craft an amazing agent experience. Now, we've had an amazing developer experience for years. But as we move into a world of AI, crafting a beautiful and seamless agent experience will be the difference between winners and losers.

0:43Joining me for this discussion is Matt Billman, co -founder and CEO Natlify, where under his leadership, Natlify has become one of the fastest -growing platforms for modern web development. He's also coined the term agent experience. But before we dive into the show's day, Are you struggling to beat model benchmarks or implement Gen AI in your product? If so, you need churring. Churring is an AGI infrastructure company backed by incredible investors like Foundation Capital and Westbridge Capital. And they do two things. Number one, they help leading companies in AI labs like Salesforce and Thropic and Meta enhance their LLMs with advanced reasoning, coding, multilinguality, multi -modality and more.

1:252. They combine human and artificial intelligence expertise to deploy cutting edge AI systems for awesome companies like Rivian and Reddit. Right now, Turing offers a free, 5 minute self -assessment to help you pinpoint your place in the Gen AI journey, get tailored next steps to optimize your model's strategy, and then finally learn how Turing can refine and implement your models for better performance. Take the guesswork out of Genai. Visit churring .com forward slash 20VC to start your free assessment today. And talking about scaling seamlessly, let me tell you about Work OS. So Work OS is the modern identity platform for B2B SaaS, helping startups move up market with ease.

2:10Selling to enterprises means honestly really complex security requirements like Samo, like single sign on, like SCIM, provisioning, audit logs. Honestly, a ton of stuff that is just a nightmare to do. Features that take months to build and maintain. Well work out a streamlines this with flexible, easy to use APIs that may enterprise readiness quick and painless. Its sweeter features include auth kit, a complete user management solution, free, up to a million monthly users, with built in MFA, R -Back, Bot Protection and user impersonation. Enterprise SSO, my god these enterprises love their acronyms, supports any identity provider using SAML or OIDC, while directory sync enables seamless user provisioning and D provisioning, oh no, for SCIN compliant directories, find grained authorization powers, compliance permissioning and it comes again, the admin portal simplifies SSO and SCIN onboarding for IT teams.

3:14Trusted by companies like Cursor. Cursor users, if Cursor users, it's gotta be great. Come on, perplexity, versel, Sierra. I mean, these companies are awesome. And they've raised $95 million in funding from amazing people like 20 VC. Try it today at workos .com forward slash 20 VC. Now that you've nailed enterprise features, let's talk about creating amazing product experiences. Get your users to do what you want them to do. That's the simple power of Pando, The only all -in -one product experience platform, Pendo combines analytics, in -app guidance, session replay, feedback management, and road mapping, all purpose -built to work together seamlessly.

3:53Trusted by over 10 ,000 companies, Pendo is transforming how businesses understand and engage their users. Plus, they're the creators of mind the product. The world's largest product management community, it's awesome. See the magic for yourself, visit pendo .io, forward slash 20 product, iPhone podcast, to get started today. You have now arrived at your destination. Matt, I'm really excited for this. I really like your writing and I've loved your writing for a while. So thank you so much for joining me. Thanks, it's great being here. I would love to stop. How does building for agents, not humans, change the way that we build?

4:31Yeah, 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 the agents just kind of escaped that they're just off doing their own things and we are 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 an where agents are fundamentally always doing something because some human is trying to accomplish something through an agent.

5:05And now we are in the very early days of building a discipline around the 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? Can I be direct? I mean, it's a nice one. I don't understand how does building for agents change the way that we build? Let me take some examples for stuff that 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.

5:52And 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 code in agent, they might just ask that agent, build me a web app with a suitable game, whatever. And now I shared with people on a URL. So we gave those agents the ability to just deploy to Netlify, no sign up from the user, So show the user the URL and say like here's what is built for you.

6:28And then later the user can go claim it and say, okay, I actually want the Netlify account or I haven't Netlify account added here. 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 their authentication space, in the database space and so on, to make their tools available to agents without an explicit users sign up and set up step first and delaying that step to after. Right? Like so that's like one very specific example, but one that's been really impactful. Does designing for agents in this way reduce the importance of UX, of design, of taste one building?

7:09I don't actually think so. In the end, we're designing for humans. I just don't think anyone can have a great developer experience five years from now if they don't have a great agent experience. So that also means that a lot of the traditional UX disciplines and so on and understanding not just the UI design but the UX design, right? Like the holistic experience of a user trying to do something with your product that's still really important. But obviously a lot of the details are going to change. A lot of things we do today are going to require very different interaction flows and very different ways of displaying that UI to people.

7:45If I'm a founder today, how do I tell my product team where they should focus, what they should build, what they should do? If I'm setting strategy for them, what the fuck do I tell them, Matt? You embrace it in the experience as a discipline in the same way that you would do it for a developer experience or for user experience, right? Like, I think we'll see that being the approach, right? Like, rather than going from like some theory of like, we need this spec now, or start understanding in your space, what are the agents that humans are actually going to be using with your product? In our space, it's on the one hand, tools like cursor, copilot, wind serve.

8:24On the other hand, tools like bold, lovable, critics, YC, refine, like all of these tools, like one level up. But in other spaces, that might be a different set of agents. When I interview people at Stemart, the common public company short in Salesforce for a number of different reasons, but do agents merely turn Salesforce into a data repository? That's definitely the big three salesforce, right? Like to just be reduced to some kind of system of record. I'm sort of mixed onweller long -term, that's a really, really bad thing 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, like all of these tools rely on, that's not like, I don't think you should underestimate to me just how valuable a position that is because all those flows and apps and everything will run on top of it and sure it might change a lot of what Salesforce looked like today and to what they look like in the future.

9:20But 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 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 UI, custom applications is going to dive down dramatically. And typically that will mean that demand for doing that will go up right, and people will start building much more custom applications for everything, right? And then the question is like, where does all that data live that those applications need to interact with and where do the rules that govern that data to make sure that it's also safe for these agents to do stuff to it, where does the authentication and permission and auditing layer live and so on.

10:11I don't think it's as binary as that. I think you can retain Salesforce as your system of record, but then also build applications around it, use other SaaS providers around it. You said before, generic SaaS, and I'm quoting here, generic SaaS tools will increasingly be replaced by custom developed internal apps. I'd love to debate this one with you, because if so, I'm going to be out of a job. Why do you think that? I think it's a balance that will really start tipping, right? Like where today, if you're a company and you have a specific business need, hiring a development team to build, deploy, operate an internal application just for you to solve that business need.

10:51The math of that very rarely makes sense, right? Like the in the build versus buy equation, the buy equation almost always makes sense, right? And that balance is going to start really shifting towards like building is going to be much more for viable option. And that's not a generic sourced tool. That's a specific tool that 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 lovable of the world. But you said generic SaaS tools and we have companies like Cloner who are saying, oh, we're replacing all of our SaaS with AI.

11:31They're not. They're absolutely not. And they won't. Why do you think they will? Again, I think it will be more and more common to start doing that, right? 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 tens of thousands of businesses. So we'll build like the generic version that works across 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 into on, right?

12:07But you typically also have to adapt how you work as a business to the tool you just bought, right? Like you typically sort of have to with a lot of the SaaS tools that I bought over time 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 we're gonna 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.

12:39It'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 leaning into a generic solution for tens of thousands of companies. You sold to large enterprises before. They moved slowly, they're not always usually educated. Respectfully, I think it is a very small customer segment that we'll be doing that in the next 15 years. No, I honestly, I probably agreed with that a year ago, and then I started seeing just how fast tools like Bolt and Loveable into an are starting to change things at sort of the other end up indie startups, everything like that.

13:18And now my timeline is more like five years and 15 years before it's going to be really relevant. It'll start by tapping into areas like everywhere where you have no code tools right now, retool and stuff like that. Starting to have agents that use code together with people to build there will start just giving large more flexibility. You'll be able to build much more. And as that will, we'll start seeing that that spread into more and more, I think like in internal tooling teams that solves like where some budget will gradually start shifting from from the SaaS spent budget and to that you know internal tooling team budget or or to kind of like external providers that custom build for you kind of budget.

14:04How do you think about maintenance? The joys of having independent providers is that maintained, updated continuously by great teams. When you have 200 tools across a large organization, you're not going to do maintenance security updates on all of them. How do you think about that? Yeah, that's something I think a lot about because the interesting thing is that that's kind of like this whole sort of gem stack architecture I proposed back in the beginning of Natalie Feister and you start becoming really important for this as well. right? Like because if you start going from buying 200 test tools to running 200 internal specializer, maybe 400 internal specialized applications, you have to do that on a platform that zero operations, first of all, right?

14:49Like, and that means that you start having this new stack where people like you can can can make a lot of money by investing in the right places. It's all of the components that go into that kind of like zero operations and serverless stack of building blocks, right? Like you definitely don't want a code chain agent that you're using to solve one of these problems to give you like a massive Kubernetes cluster and be like, good luck. You'll want it to deploy to platforms that are zero operations and serverless and with like security built in and so on. you'll want them to use serverless API services databases that are all managed with their own security team and operations teams and so on as the as the Legos it kind of builds on top with and composes together because otherwise you're right, right?

15:41Like otherwise if it's just fully custom build, Kubernetes cluster stuff, all of it, the viability of that whole approach goes away, right? Like the maintenance cost just overtakes everything. And then you could speculate can the agents maintain it, but I think that then we are definitely in a 15, 20 year horizon before they could be relied on for anything like that. But if the agents build on top of a stack where your deployment layer is fully sewer operation, serverless and so on, where you're using systems for headless content or you're using Salesforce as your system as records for a bunch of stuff, right?

16:20Like you're using managed databases, whether it becomes like a super base or a knee under the ear or some of the big cloud providers that wins out in that layer, right? Like, but the agent used these building blocks that are in this in the self maintained by someone, right? Like, then you start being able to take on that level of building without exploding in terms of like maintenance costs. When you look at the different stages, we mentioned that lovable and, you and being able to create very easily. Prototyping is now like step one. Have an idea. Does design tools like Figma die in the next wave of build?

16:58They're definitely gonna have to really think a lot about this full stream of how prototyping again is often gonna shift from the design tool into building something. I still think for a certain level, again, we talked earlier about like this brand matter, just 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 FICMAD doesn't manage to connect that to the fall into 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 and 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?

18:03Like to prompting as one of the key ways to just start a new project. 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 to say the worst thing with charity is it made chat and prompting the default interface. Do you agree? Yes, I don't think chat is like the end of all of interfaces. It's a very simple, transactional flow that's easy to work with and I think chat or conversational interfaces will be a thing. It's also a thing between humans. We all use Slack and Messenger and WhatsApp and all of these chat tools for good reasons.

18:41It'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 to do in the AI powered world. So 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? 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 experience as that we've built through the last 10 years, right?

19:24Like, and we're 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 a LLMson and AI available that we couldn't build two to three years ago because I think there's a lot of interactive user experiences that we can now build that are definitely not just a chat prompt saying like type what you want to see a do or like talk to me right like that could not be done if we didn't have the ability to one create all kinds of assets from 3D modeling to sound and music and visuals and so on automatically.

20:10And that couldn't be done at a time when the core constraint of all user interfaces is that computers are transactional deterministic machines. You click a button, the UI does something that you know that button will do every time. You type a command, the computer does something predictable. right? Like, Jennie, I kind of removed that underlying constraint and we haven't really started building new kind of UI that takes advantage of that. And I really agree that like, Chad is, Chad is not yet there, right? Like, Chad is part of it. But like, if I, if I follow a link and I land on a e -commerce from like a large brand that's trying to really differentiate themselves and be something different, right?

20:52Like, I definitely just, I wouldn't want just to have some box saying, like, do you come up with what you want me to do? It's like type something, right? Like, I want, I want those online experiences to guide me and to have some experience for me, right? Like, and to take me somewhere. And that's never going to happen through chat style prompting. Why? Do you not go into a shop and ask what you want? Yeah, but you also have like, you think about how much a fancy brand spend on that shop, right? Like, and on that, this, on the display windows 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 I have a look at this thing?

21:32Right? Like that's a very expensive industry. I like 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 storytelling, you also got to find ways to do that. Today, honestly, like most decomperses, for example, like the last three years have kind of been very stagnating in terms of development in that field. Everything looks pretty similar with some changes. But I think when the cost of building and e -commerce that looks like any of the current brands today goes down to ask both to do it for you, ask levelable to do it for you, and you have that, those brands are going to have to come up with something much more engaging if they want to keep stand out and if they keep wanting to build relations to people less, more than just like, one click shopping experience amongst it any other.

22:25You've mentioned Bolson lovable and when we think about these Gen AI services that are scaling revenues, Matt, like we've never seen revenue scaling before. I don't know. Everyone says these Gen AI tools, they're sugar high revenues. You can't rely on them. How do you distinguish between sugar high revenues that lack sustainability and meaningful durable revenues that can be relied upon? When I look at tools like like bold and lovable, the category itself has something where it seems obvious that there will be a lot of money spent there. Just because like when you see some of the examples of what people can do with those tools that they just couldn't do before, it seems obvious that that's worth something.

23:10When you see how fast someone can build a typical website or typical app or dashboard, right? It's so funny. This is the promise of Webflow and Squarespace for years. Anyone who can build it, the truth is, Shopify Webflow Squarespace, me and you both know, my mother can't build on them. It's still too tough. She can't use Lovable. Yeah. But even then, I don't know. I've like this example of this UX designer from Neantic that are building these things on a fault that just seems, I know that it's just mind -blowing, right? Like, he built this whole 3D object editor, like this actual tool where you can build 3D objects in the browser with AI -based prompting and everything running in the browser.

23:54And you build that entirely with Bolt just by prompting. And that's so mind -blowing, but it's also an example. I'm sure your mom probably couldn't do that with Bolt, right? Because I don't think I could do that. I'm pretty sure I couldn't do that with Bolt, right? like that takes like real skill having built that skill set to work with these tools. So I'm sure we'll see the same that not everybody is going to use these tools to make their self stuff in the end. Even if you can potentially as a small restaurant owner or whatever go to Squarespace today and build yourself the website for your restaurant, you're probably busy enough that what Squarespace will do for you is making sure that the cost of getting some freelancer to build your website is much lower than it would be if you had to get that freelancer to program the whole side from scratch, right?

24:45And I think we'll see the same with tools like Bold and Lovable. I think we'll see the next 100 million software developers emerge around tools that are authentic, but I still think it will be a profession to use those tools to do stuff, to build stuff for other people because they'll still be a skillset involved in knowing what the agent should actually build for someone and what works and how to get their fastest and how to push the limits of these tools. But it will lower the cost of each project built with those tools dramatically from what it would cost today. So there are devs using cursor, there are devs using loveable and bolts, and then there are non -devs using your love goals and your bolts.

25:30Yeah. When you look across that stack, how many do you think remain in the devs using cursor and your alternatives? Because I've been in this space of building tools for web developers for way too long, I guess. I've kind of seen the same discussion forever, 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? And I was as CTO building the whole platform that our freelance designers would use to do that work on and everything, right? I can 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?

26:13Like 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 cam stack movement around decoupling the actual web front end from from the back end and starting to build at that level, right? Like and we saw probably in the years since like 2016, 17, 18, we started seeing all these boot camps just like starting to spit out react developers. And there was like the sentiment that they're not real developers and they don't know the whole stack and so on, right?

26:50Like, But it turned out that they were really valuable and that working at that layer of the stack was important and now we might have like an additional like 14 million Developers in the world from from that layer, but the interesting thing is that the layer below of like Ruby and Rails Developers and Java developers and dot net developers. It's probably still at least as large as this was before that the new layer Just got added on top and my prediction is that something pretty similar will happen, right? like we'll probably add and you in this case I think more like a hundred million new software developers that mainly operate at this like very agentic stack and some of them will go a couple of layers just like now right like you'll have some that are more full stack that others that'll go a few layers below that level of the stack will have that happen and then you'll probably have more or less the same amount of developers that are actually like fully understand the layers below when I'm working on that and I'm doing work on that.

27:48And those are probably the ones that are going to be thinking a lot about the agent experience of what they're building, but they're probably still going to be really, really necessary. I totally get you. When you look forward to that future, OpenVers is closed is a very big question. How do you think about which approach dominates in an agent first world and why? Yeah, I mean, I think every time there is a platform 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 that once that were best positioned initially to become fully vertical lockdown platforms was really like the large language model makers.

28:29And 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 caused kind of went down and like this open model came around and deep sea 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 Is best than 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 which whichever model is best.