“I haven’t written a single line of front-end code in 3 months”: How Notion’s design team uses Claude Code to prototype

[00:00] The way I think about designing B2B SaaS is you want your designs to encounter reality as early as possible. I've always been into prototyping and then all of a sudden these AI coding tools come along and now I can prototype faster. I can prototype in production. So explain to us what this prototype playground is. It's just the Next.js app. All of our [00:20] Prototypes are in one place. Seeing what other people are working on is really fun and interesting. And oftentimes you spot cool ideas and you're like, "Ooh, I wanna try that." The code is all in one place. It's just in one repo. And so I can just yoink, [00:31] cool ideas from other people's prototypes and incorporate them into mine. Every time somebody is like a little anti AI assisted coding, I'm like, do you know that I used to have to walk uphill both ways for my CSS? It was not fun to do this. I mean, even just sitting here watching this, I still just find this magical. [00:52] Welcome back to How I AI. I'm Claire Veau, product leader and AI obsessive here on a mission to help you build better with these new tools. [00:59] Today we have a designer-centric episode with Brian Lovin [01:04] designer at Notion AI, who's going to show us how he set up a prototype playground for the entire Notion design team to vibe code using cloud code, [01:15] any prototype they need. [01:17] This is a great one for someone looking to either shift their design organization into a code first prototyping mode or learn some advanced techniques with Cloud Code. Let's get to it. [01:28] This episode is brought to you by WorkOS. AI has already changed how we work. Tools are helping teams write better code, analyze customer data, and even handle support tickets automatically.

[01:39] But there's a catch. [01:41] These tools only work well when they have deep access to company systems. Your copilot needs to see your entire code base. Your chatbot needs to search across internal docs. And for enterprise buyers, that raises serious security concerns. That's why these apps face intense IT scrutiny from day one. To pass, they need secure authentication, access controls, audit logs, the whole suite of enterprise features. [02:06] Building all that from scratch? It's a massive lift. That's where WorkOS comes in. WorkOS gives you drop-in APIs for enterprise features, so your app can become enterprise-ready and scale up market faster. Think of it like Stripe for enterprise features. OpenAI, Perplexity, and Cursor are already using WorkOS to move faster and meet enterprise demands. Join them [02:36] day. [02:38] Brian, welcome to How I AI. What I am so excited about in terms of our conversation today is, you're going to show us about how one of the best designed products out there, Notion, [02:51] is being designed by people like you using these new AI tools like Claude Code. So why did you make this shift to how you were doing design, what it meant to prototype design and build things, especially... [03:06] for a product and in a company who values design so highly. [03:09] The way I think about designing B2B SaaS is

[03:13] you want your designs to encounter [03:16] reality as early as possible. [03:19] And, you know, if you imagine this gradient of like I'm scribbling on a napkin on one side, [03:24] to [03:25] I'm shipping to production and showing customers on the other side. [03:28] Our goal as designers is to move [03:31] up that gradient [03:33] towards prod as quickly as possible. [03:36] So I'd say for most of my career, I've sort of biased towards being interested in programming, mostly at the prototyping level. I just find that when you're... [03:47] designing something in Figma and then you actually try it in the browser and [03:50] in the browser you notice a ton of problems. [03:53] you know all of a sudden you're clicking things you notice loading states you notice ah that didn't quite work on this screen size so you encounter some version of reality sooner and you end up getting to a better design [04:05] more quickly. [04:06] So... [04:07] I've always... [04:08] been into prototyping and then all of a sudden these AI coding tools come along [04:12] And now I can prototype faster. I can prototype in production. [04:16] I can, or what I most often do now at Notion is just prototype in a little internal tool we've built called Prototype Playground. [04:25] and [04:26] again the idea is just like how do we get something that's somewhat realistic [04:31] in a kind of real environment, in our case, the browser, as quickly as possible. [04:37] And I think that just helps you move faster and end up with better designs. [04:42] So explain to us what this prototype playground is and how you set it up and how you might use it.

[04:47] Okay, so Prototype Playground is [04:50] Nothing magical. It is just a next JS project. So I actually hear it to source apps and [04:57] And... [04:59] There's an app directory. And you'll notice here in our app directory, we're normally in a Next.js [05:05] uh, [05:06] app you would see pages. Well we've just namespaced every designer on the team or PM or engineer [05:13] whoever signs up and wants to use it. [05:15] can just namespace some directory. So here's Brian, and then every directory inside of that is [05:21] some prototype. [05:23] And so it's just the Next.js app, but each page is sort of a standalone [05:28] There's no global layout. There's no global, I don't know, like structure that you have to adhere to. [05:34] And so what that looks like on the front end is this. This is what we call prototype playground. And it's just a list of prototypes ordered by who is working on stuff. [05:44] recently so here's a few from december and then a bunch from november and it's really cool because having everybody's prototypes in one place is useful in two dimensions one just from a visibility point of view seeing what other people are working on is really fun and interesting and oftentimes you spot cool ideas and you're like oh i want to try that [06:03] And then on the other dimensions, like if you spot a cool idea and you want to try it, the code is all in one place. It's just in one repo. And so I can just yoink. [06:11] cool ideas from other people's prototypes and incorporate them into mine. [06:15] Usually by just telling Claude to do that.

[06:18] uh, [06:20] Yeah, I think before Prototype Playground, there was a lot of designers at Notion who prototyped in code. [06:25] The difference was we were all creating our own [06:29] repository our own next js instance and so it was hard to know where everyone's stuff was everyone was rebuilding it in different ways [06:37] or if people were trying to recreate something that looked notion-y, we were all doing it from scratch. [06:42] So anyways, Prototype Playground, Next.js app, all of our prototypes are in one place. [06:47] And then we have a few shared components and shared styles. [06:51] So if you want to make something that looks Notion-y, you can do that pretty quickly. So for example, we have some templates here I can show you like [07:00] Notion UI is just a notion a sidebar and actually this isn't even very notion II I think at some point I slipped this new button in here, which obviously doesn't exist in the past [07:08] the product. I don't think these things do anything. But it's close enough. If you're like, oh, I needed to prototype something with an ocean sidebar, I can just come in here and duplicate this template. And then we, of course, pull in a bunch of our colors, typography, icons, so that [07:25] again just getting to close enough notion styles [07:29] without a whole lot of effort. [07:31] Yeah, and I want to call out for people, a couple many episodes ago, I showed how you could build a very similar Next.js app for yourself that had a combination of docs you were working on, markdown docs you were working on, and prototypes in a very similar format where it was like, here's my folder of just stuff I'm working on.

[07:50] very minimal shared components, very minimal shared styles. I like this, too, because it's nice to have that team level organization. So you can pop in and see. [08:00] who your teammates are working with. I have a question from an operational perspective. Did you set this up? Was this like a passion project for you? [08:07] Did engineering set it up for you? Like how did this actually get created? [08:12] Yeah, I set it up with another engineer. I mean, it's just the Next.js app, but then operationally, just a few approvals. It's deployed on Vercel, so we had to go through a little bit of [08:21] process to get that project spun up. [08:23] get a few people added as members, [08:26] Otherwise, yeah, it's not that much. Again, it's just a pretty basic Next.js app. [08:32] which you can literally use Claude to like [08:35] help me make a next js app and it's just going to get you the the default i like you know i like keyboard hands everybody does the same keyboard hands motion where it's just this i have one more question which is yeah of the people now working in this repository how many before [08:50] were working in [08:52] code versus this is their first repository that they've cloned [08:57] to, you know, their desktop or deployed. Was the design team pretty technically adept already? So this was very natural? Or were there some people that needed to be onboarded? [09:06] I think so. I mean, to be honest, Prototype Playground is still [09:10] Really for me like I think I use it the most you can see here There's a bunch of other people that that are creating things, but if you were to go through I probably use the most I think I

[09:21] There's maybe like [09:22] five to ten people at notion that use it quite a lot and then a bunch of people who [09:29] either have tried it and it didn't stick, and we can get into reasons why that is, or they're just not interested, or they prototype separately, right? Like we still have people prototyping in Figma. We have people... [09:40] that prototype in their own code base still, they just prefer their own stack. Maybe they don't like next JS. Maybe they don't like react. [09:46] So they do something else. [09:48] And I think all that is totally fine. In fact, one of the features I added recently was this ability to link to an external prototype. So if you prefer using v0 or lovable or a Figma make file, whatever it might be, [10:01] You could just link to it here, and in fact, this is what it'll show up as in Prototype Playground. Just have this little external icon. And so you can click it, and it'll open in a new tab. [10:10] So in theory, this could be the prototype playground or repo for [10:16] any prototyping tool, [10:17] My hope is that over time, [10:20] we make this thing [10:22] useful enough that more people will want to prototype in it, [10:25] because it's just faster than those other tools [10:28] And we got to figure out how to lower the... [10:31] onboarding complexity for people who aren't technical before so to answer your question um [10:37] I don't know, I'd say some people who weren't technical [10:41] made their very first code prototypes or like ai assisted prototypes in the playground [10:46] But probably the majority of us that are still using it daily had some technical background.

[10:52] Got it. Perfect. Well, let's prototype something. [10:55] I want to see how it actually works. [10:57] Let's do it. Okay, so... [11:00] there's a few ways to make new things in next.js right like we could be in cursor and we could [11:05] come in here and create a new folder and create a bunch of page.tsx and metadata files and [11:12] that sucks i don't want to do it um so there's two ways around that the first is [11:17] When you're running in localhost you can actually just click this button that says new. [11:21] and you give your prototype a name and a description, [11:24] I'll call this one HowIAI. [11:27] And then this is for fun. [11:29] and i create that and all that's doing under the hood if we bounce back over to cursor [11:34] is it just created those files for me on my computer. This is my favorite part is like, there's no backend for prototype playground. It's just all... [11:42] Files on disk and then we can just push all this to get hub. So here we have like a little metadata file and [11:48] These get sort of collated to render the list on the home page. [11:54] we have an actual [11:55] uh... [11:56] prototype file here with some code and then [11:59] This is kind of nice, like it automatically gives you a button to open it in cursor. So now I can just come in here and start prototyping. [12:06] Now, typically my workflow is [12:08] I just bust open Claude in the terminal. I know this isn't how you're supposed to use cursor, but it's just... [12:15] It's just how I do it. It's probably not even how you're supposed to use cloud code, but I just do it. We're just equal opportunity offending these two tools. I know. I know. Sorry, everybody. But this is how I like to work.

[12:29] And in fact, I have a little shortcut here where I can just [12:32] Press, uh, [12:33] Caps Lock G. [12:35] and then I can get these two things side by side in my computer site usually. [12:41] Clotting over here, reviewing the changes here, and then monitoring sort of the output over here. [12:48] So let's see here. [12:51] i want to make a prototype and i don't know let's just come up with some contrived example like maybe you can help me think of a [12:57] All right, good. [12:58] Good use case. Can we make a prototype for... [13:03] Oh, like a little video and audio. This may be complicated. A video and audio display module for my podcasts. [13:11] Video and audio. So it's like video and then maybe like an audio player. Let's see. You know, it's it's opus four or five. I think you could do it. [13:18] okay let's try it so normally uh let's walk through like my actual workflow there's sort of two steps one is you can type a lot that's not that fun i do use this tool called monologue where you can just talk to your computer there's many [13:31] products like this. I think Monologue is just nice and cute. So we can just talk and it's just much faster than typing our prompt. The second thing you'll notice with Cloud Code is I switched over to plan mode. [13:41] I think it's really, really important to plan before [13:44] doing anything for whatever reason you just get better outcomes [13:48] Now, the key thing about using plan mode is... [13:52] to actually read the plan and i think this is where [13:55] having a [13:57] Development background just gives you an edge.

[13:59] Because you can read the plan and be like, oh, that part actually doesn't look quite right. Whereas if you maybe don't have as much programming experience, it would be harder to tell that. [14:08] But in either case, I still find that having the plan mode and creating some structure before actually writing code [14:14] is better so let's just do both of these things at the same time so we're in plan mode [14:19] And I'm going to [14:21] invoke monologue here and it's recording and so let's say [14:24] I want to build a new prototype in this how I AI directory. [14:30] And we are a podcast... [14:34] and I want to [14:36] build a detail page for a podcast episode that has both a video player and an audio player underneath, [14:43] the page should have [14:45] the title of the episode, description, [14:48] And how about [14:50] If you... [14:52] Hit. [14:53] play [14:54] there's little confetti that shoots up out of the player. And so we... [14:58] And that, and now I will [15:01] Delete this. [15:03] And we plan. [15:04] Thank you. [15:05] So I have to give you props on two things. One, I am also a plan mode slash like write your spec, write your PRD person. Obviously, I think the second thing is I am still just such a read the code, read the outputs. [15:20] girl when it comes to AI. It's actually one of my challenges when I use something like [15:24] Claude code or watch people use Claude code is if you don't do it inside a cursor, [15:29] or something that gives you this sort of, I love your three pane window, your code window, your clawed window, your output window. Because I see people with like 17 tabs of clawed code going, just accepting a bunch of changes. And I have to read. I think this is also just...

[15:46] like engineering development background where you can just spot things that make [15:50] no sense in the moment as opposed to having to go back and debug something. So I am very much aligned with you on that. [15:58] Yeah, it's helpful. And you know, this is probably obvious to a lot of people who are familiar with using cloud code, but maybe if you aren't like another piece that's really important here is [16:11] getting the right context up front, right? Like we just [16:14] typed in some prompts um [16:17] But under the herd, I can show you we actually have [16:20] Some other files helping us out here. So we have a clod.md file at the root of our project With just some rough instructions around like the tooling that we use like we use bun we use tailwind it has like a rough outline of the project structure and [16:35] Another thing that we do is anytime someone runs the project locally, [16:40] we create a clod.local.md file. And that local file is not committed to the Git repo, so it's personal per computer. [16:49] And it adds a little bit of extra context, like, hey, this is my username in Prototype Playground. It tells... [16:55] Claude, where my directory is. And it gives some instructions like, hey, you know, don't go around and mess with other people's files. Like, prefer to work in my directory. [17:05] and a little bit more about the workspace structure and how like individual projects can be built so a couple of those things are working under the hood here [17:14] And while you're accepting some of these Claude code changes and questions, I do want to call this out for folks because I think people are pretty aware of the Claude MD.

[17:23] global settings, but I think people forget that there are actually [17:28] locally scoped versions of these that you can implement. And so [17:32] it's really useful to get one version deployed to everybody that gives you your master rules for using Claude and then [17:39] you can set up your own custom one with your own particular preferences. And I think that's a really nice way to create a good collaborative environment where people are using a similar AI tool or agent to work in the repo. [17:53] Totally, yeah. [17:55] Okay, I don't know. We'll see how this goes, but it's gonna install some sort of confetti. It's gonna have a player audio. This is really awesome. Like it does a wireframe in the plan, which is crazy. [18:06] Um, [18:07] And here, I don't know, we can just kind of skim this for the sake of this example here. [18:12] Uh, this looks fine, so let's auto accept edits. [18:15] Now, I have a tip for people because I think [18:20] When you spend enough time on Twitter or watching other people use these coding tools, [18:25] People are always like, how do you get it to run? [18:28] for longer. [18:30] You know, they find themselves constantly getting stuck or the agent does the wrong thing or it's asking for their input. [18:37] And... [18:38] My philosophy on this has been [18:41] any time the... [18:43] ai asks you to do something [18:46] you should, before responding, [18:49] Try your best to see if [18:51] you could teach the AI to answer that question for itself.

[18:55] There's a good example. Oh, wow. That was very fast. [18:58] - Ooh la la. - Well here, let's hold on that and see if the confetti works. Well actually here, the example is I've already taught Claude to like always lint itself after it's done, right? Like what's really annoying is when it builds a bunch of stuff and then you go and look in your browser and there's some error, right? [19:15] So for example, I've taught Claude [19:17] hey, check your work. One, you can run commands like, what was this? Like ESLint, right? And like look for actual TypeScript errors. The second is you can give it access to MCP tools. So Playwright is one, the Chrome DevTools MCP is another one. [19:31] And you can say... [19:33] Well, actually, before installing this, [19:36] Claude would say to you, [19:38] "Hey, I've implemented your feature." [19:40] Go take a look at it and let me know what you think. [19:43] Um, [19:43] And remember, our rule is anytime Claude tells you to do something, ask if you can teach it to do that thing for itself. So I don't want to have to look at the browser every time to see if I did it correctly. [19:54] So instead I... [19:55] teach Claude actually you should be the one to go and open the browser so it knows how to launch Chrome it knows how to navigate here and knows how to click the play button look for confetti make sure the audio is working [20:06] all that kind of stuff. [20:07] And so now we were able to run this task for much longer without my input. [20:12] and actually get to something that [20:15] is working well. Um, [20:17] I'm actually very impressed with this prototype. It's much more lovely than I thought it was going to be. [20:23] to end up much more robust. And the confetti looks great. - The confetti looks great, yeah. Well here, I'll show you another example. This is, I think, where the power of MCP gets crazy. So let's actually,

[20:34] Clear this we're just gonna start [20:36] start a new... [20:37] new conversation here. I'm going to... [20:40] Just totally undo everything. [20:42] Let's just start from scratch. [20:43] So, uh, [20:45] A couple other things that I've built in, you know, I think, uh, [20:50] remember, like, I'm trying to [20:51] make the [20:53] onboarding flow as simple as possible for people on my team. [20:57] So, [20:59] What Claude has is called slash commands. And you can just build these yourselves, and they're basically glorified prompts, but they can also run scripts. And so we have some slash commands in the project that help people get going really quickly. So I have one called create prototype. [21:13] And then you can give it an optional name, so we'll call this one [21:17] how I AI [21:18] And that's gonna do the same thing as clicking the new button. [21:21] on [21:22] on the browser, which is what we did earlier. The difference, of course, is [21:27] I don't have to click things. I kind of want to [21:29] designed this so that I basically live over in the terminal. And can you show us really quickly in your repo just how these these commands get defined? Perfect. Thank you. Yeah, sure. So [21:41] Uh, [21:42] Again, it's basically a glorified prompt. It has a name, a description, and then some instructions. [21:50] So in our case, we say, [21:52] uh, [21:52] kind of how to come up with a name based on what the user provides, [21:57] tells it where to look to determine the current user's username, how to create the new thing, [22:02] It actually provides some sample code to use for both creating the page and the metadata file.

[22:09] I think I need to... [22:12] Also approve this so it goes. [22:15] Let's just do... [22:16] Blank for now. [22:19] as well as creating the metadata. So, you know, [22:22] AI... [22:23] is [22:24] better with good context but it's also really really good if you just provided examples of how to do things [22:31] So the reason it's important to provide these code snippets is to show it what success looks like, right? Yep. If this was like [22:39] just instructions to create blank files it wouldn't know what to create so [22:43] In our case, we're just showing it an example of success. [22:47] And we could probably simplify this. It's actually quite a long command, but... [22:51] But here we go. So it created this and a blank [22:54] piece of text. That's great. [22:56] Thank you. [22:57] So that's just one way to start. You just type slash create a prototype, and then that'll create. But maybe we have some design in Figma. [23:07] and we want to build this [23:10] This... [23:11] Might not work, but let's try it. [23:13] So we can connect to the figma MCP and I can just copy a link to this frame and say like let's build a [23:20] this, uh, [23:22] this notion UI. [23:24] So, [23:25] uh, [23:25] Before you could just paste [23:27] a link to a Figma URL and try and manually invoke the Figma MCP, [23:33] And it would sometimes ask clarifying questions, and sometimes it would [23:36] build it and then sort of stop halfway through. [23:39] i don't like any of that so we actually built a command

[23:42] called slash Figma. [23:44] And it roughly does a couple of things. The first is it actually checks that you have the MCP server installed and running. [23:52] you know, for people on the team who have never... [23:55] done mcp stuff before they might not know how to do this [23:59] And so it detects if you have it installed and if it doesn't [24:02] if it finds that it's not installed, it'll just teach you how to do it. So it actually returns instructions to the user on how to set all this stuff up. [24:09] And then it moves on to phase A, designing [24:12] uh, [24:13] or extracting the design from Figma, then it will implement it. And then the most important thing is we enter this third phase called the verification loop, [24:21] where it's going to open the browser and compare the implementation it created to the original Figma file. [24:27] And I think my instructions are basically [24:30] keep looping until you've gone through like two loops where there were no more changes oh yeah here repeat until the implementation matches or after three iterations with no changes and [24:41] and then stop iterating so let's just see what happens this [24:44] I would say it gets like [24:47] 80% correct 80% of the time, but that's just how AI is right now. I was going to say about 60%. So I think I was right. [24:56] Well, actually, you know, I think it is 60%, but this command and this loop and these instructions and like the pairing of the two MCPs actually gets us... [25:05] to 80%. [25:06] I want to call this out for folks because one of the things that I find most frustrating using MCPs even as a

[25:13] fairly sophisticated user is one [25:16] You just have to use these like magic keywords to invoke the MCP and the right tool and the right thing. And, you know, sometimes I have one of the challenges I have is I have a lot of MCPs that use the same tool. [25:28] tool names because so much across SAS is named the same. Like everybody has the concept of projects. Everybody has the concepts of pages or documents. And so I like this idea of like force invoking a specific MCP [25:41] via a slash command and not even just force invoking that specific mcp [25:46] but force invoking a specific set of tools in that MCP. [25:50] Super, super useful. And then I will give you props for the instructions. [25:55] at the top that teach somebody if you have no idea what you're doing here, how do you even get this thing installed? That's such a nice piece to add in as user experience for [26:05] a consumer of this slash command that might not be you. [26:10] And so that's something that people should really, really think about. [26:13] Yeah, yeah. I would say also it's funny because I've actually watched a bunch of these videos and looking even back at the ones from six months ago, [26:23] It's crazy how far the tooling has come. [26:26] And so I imagine that people who, for whatever reason, might be watching this video in six months will look at what we're doing here and be like, oh, how naive. We've come so far. MCP is no longer a thing or something like that. Right. [26:38] And I kind of feel that way now where MCP is...

[26:43] It's like, [26:45] Not the best thing. [26:48] But it's the best we have so far, right? Like it's very context inefficient. Sometimes it runs forever. Sometimes it just like blows up your context window. [26:57] But it's the best we have right now. So even just watching this, right? Like here's our design that got built. [27:02] This was literally just pasting the link to the Figma file. [27:06] no other custom instructions, and now over here on the right it should be [27:10] Uh... [27:11] I think I ran into an issue earlier. Yeah. [27:14] Something got busted with this. Let's [27:16] Try the Chrome. [27:18] Thank you. [27:19] DevTools, MCP again. [27:22] I think I quit it midway through because it was detecting some conflict with the window. Anyways, this is pretty good by default. And then from here, I would iterate some things you might notice would be like there's no hover states. Some of these images are broken, but those are just easy follow-up tasks. [27:36] Well, and you're doing this from a kind of design perspective, but think about how many engineers sit there and like, [27:44] pixel pullover Figma prototypes into the front end. And, you know, if you have a great design system, maybe that's easier to do, but it's not what the 27 seconds that we just watched to scaffold stuff out. And so, [27:59] I just think, you know, the friction reduction in these [28:04] you know asset to asset handoffs which for my entire career 20 plus years in tech have been [28:09] the most expensive part of implementing something, where a designer gives you a design,

[28:15] and then you have to get into the front end, or the front end has to be hooked up to the back end. [28:19] All those little pieces can be smoothed out and done much faster. And then you can spend the time on... [28:26] the optimizations, the performance, the how it feels, how it how it works. And I think that's just really [28:32] It's really fun from a builder perspective. [28:35] Totally. It's so fun. And yeah, I mean, even just sitting here watching this, I still just find this magical, right? [28:41] Now that it's using the Chrome DevTools MCP, it looped and fixed the broken images and created this checklist of stuff. Like, okay, everything appears to be right. It's got this bottom bar. These things are obviously wrong, but we could go and fix those with the follow-up prompt. But again, the goal is like, [28:58] Can we get 80% in literally one prompt? I just pasted a link and it just iterated itself towards something that's roughly complete. [29:05] I know. And every time somebody is like a little anti-AI assisted coding, I'm like, do you know that I used to have to walk uphill both ways for my CSS? Like, it was not fun to do this. I find this just... [29:20] mesmerizing this is so cool mesmerizing [29:23] This is great. [29:25] This episode is brought to you by Orcus, the company behind Open Source Conductor, the platform powering complex workflows and process orchestration for modern enterprise apps and agentic workflows. Legacy business process automation tools are breaking down. Siloed low-code platforms, outdated process management systems, and disconnected API management tools weren't built for today's event-driven, AI-powered, cloud-native world.

[29:55] you get a modern orchestration layer that scales with high reliability, [29:59] supports both visual and code-first development, [30:02] and brings human, AI, and systems together in real time. It's not just about tasks. It's about orchestrating everything. APIs, microservices, data pipelines, human in the loop actions, and even autonomous agents. [30:17] So build, test and debug complex workflows with ease. Add human approvals, automate backend processes and orchestrate agentic workflows at enterprise scale, all while maintaining enterprise grade security, compliance and observability. Whether you're modernizing legacy systems or scaling next gen AI driven apps, [30:38] Orcus helps you go from idea to production fast. Orcus. Orchestrate the future of work. Learn more and start building at orcus.io. That's O-R-K-E-S dot I-O. [30:54] Are there any other commands that you think are super useful? [30:58] Yeah, yeah, I can show you a couple. So, [31:01] I want to scroll back up a little ways, actually. There was this step very early on [31:07] where you can see it was running over and over again this skill called [31:11] Bunrun ClaudeSkills find icon. What's that? Well, if you look over here in our design, we actually have a bunch of very notion-specific icons, right? Like we have this AI face, we've got home, inbox. [31:24] uh we have all the icons in our project the problem is

[31:28] AI is really bad at estimating [31:31] what the name of an icon should be. [31:34] or rather it uses like the most obvious name possible, which doesn't always match what's in code. So for example, like this face icon, [31:43] There's no way. [31:45] "AI would know what we call this." Or a very common one is, [31:50] it will... [31:51] if you have like a search magnifying glass, right? [31:54] it will just assume that it's called search icon [31:57] when in fact in our code it's called magnifying glass icon. [32:01] And so this [32:02] Icon hallucination was getting really really annoying. So I wrote a little skill called find icon and [32:09] And the skill basically says, like, anytime you're going to implement an icon, [32:14] First, go and actually look through the whole project, but also look for synonyms or closely related words to the icon. So if you're gonna look up something called search icon, [32:27] Also try search for magnifying glass icon and it actually wrote a TypeScript script to do this to just iterate through all of the files in our icons directory which is like 5000 right it's a lot so it actually be very inefficient for it to try and load all that up into context it needs to write itself a script. [32:45] to do more effective searching. [32:47] So in that loop here, [32:50] Um... [32:51] yeah you can see it like [32:53] found it looked up magnifying and found the magnifying glass icon it looked up inbox and it looked up gear and trash in order to get all these things correctly

[33:02] Now this only, this skill, [33:05] had to exist after all of us on the team just got really, really frustrated with it hallucinating over and over and over again. [33:11] It's sad because it obviously missed these bottom three. It didn't get them correct. But the fact that it got these on the first pass is a huge step up. So the way I think about it is, you know, we have these commands that you run manually. [33:26] And skills are these capabilities that [33:29] the AI should detect automatically and sort of use at the appropriate time, and it'll know to do that based on [33:37] the description and title you've given it. So in this case, find icon and then how to search for icons. [33:42] And of course, the best part is just [33:44] letting it do things programmatically on your computer by calling actual [33:48] coded scripts. [33:50] This was really helpful, saves us a lot of time and just fixing imports and [33:54] Nope, search icon does not exist. Those kinds of annoying, annoying steps. [33:58] Well, what I like about this is, one, this is exactly what you would do to a junior designer or engineer onboarding. You would explain. You'd be like, sometimes we call it search, but not really. It's magnifying glass. You just got to go find the closest synonym. And the ability to be able to describe that to an agent or a skill or a tool and then let it do it programmatically for you is really useful. We do have a How I AI episode on Claude's skills. [34:28] which I think is really important, is Claude skills can be bundled with scripts. And so the ability to combine both

[34:35] you know, natural language prompting, which is in the skill dot markdown with [34:40] a set of programmatic tools in terms of scripts is a very powerful combination and [34:47] Claude's very good at making these. [34:49] Oh yeah, like all this, like I did not type a single line of code in this, right? Like this is 100% like, hey, I just need this problem to be solved, create a skill for it, and then creating that. [34:59] skill also create a script so that you can work more effectively like this is a hundred percent prompted [35:04] Show us your last command, because I think this is a really useful one. [35:08] Okay, this is fairly new. I think I merged this last week. Going back to sort of the problem with Prototype Playground, it's still a Next.js app. It's still React and TypeScript and Git and branches and... [35:22] It's just a lot of concepts to throw at someone who maybe is used to only prototyping in Figma or they're intimidated by a terminal or code. [35:31] And so I'm trying to figure out, like, how do we make this thing more approachable? How do we... [35:36] make it easier to onboard, but also not [35:39] dumbed down [35:41] Right? Like... [35:43] I want people to learn how to use computers. I want people to... [35:47] even subconsciously absorb the ideas of git and branching and pull requests and merging [35:54] So I don't know the best way to do that, but my first attempt is to create this skill called, or this command called deploy. [36:01] And, [36:03] Deploy does... [36:04] basically two things. The first is it like goes through prerequisites and makes sure that it has the GitHub CLI tool installed on your computer and that you're authenticated.

[36:14] And if you're not, it like walks you through those steps, how to do it. [36:17] And then the second step is [36:20] It will just walk you through everything. [36:23] step by step, [36:24] "How to get this prototype you've just created, [36:28] deployed so that you can share the link with someone on your team [36:31] uh, [36:32] Let's see what happens. I'm going to try it now. I'm going to hit deploy and we'll see what happens. Um, [36:38] There's a couple of really cool loops in here that I think save people a lot of time. So we can see it going through the prerequisite steps here. It's making sure I'm logged into GitHub. [36:49] Now, the first thing here, look, it's looking to see if I'm on a Git branch. It notices I'm not. I'm on main, and it shouldn't be doing that, right? We never want to push to main. So I think what it should do [37:01] is help me create a new branch. [37:05] And... [37:07] We'll see if it actually does it correctly. It's also trying to find some TypeScript errors and it's going to run some tests. [37:14] I basically told it to do all this stuff because it's really annoying if you push code to GitHub [37:18] wait for all the checks there to pass. If they fail, then you gotta come back to your computer, fix stuff. [37:23] OK, great. So it created a branch. [37:27] Now it's staging our changes. Nice branch name. [37:30] - It's perfect. [37:32] Creating the commit. [37:34] I'll give I love this. This is a great idea. I will also give my just like hack to learning get for anybody who [37:40] hasn't used it. I just love the GitHub desktop app. It just like it gives you buttons for all this. You can see your diffs. You can like create branches with buttons. So I think this is awesome. And if you are intimidated by the command line,

[37:55] there's like literally a beautifully designed desktop app that you can use. That's true. It's pretty nice. That's true. [38:01] - [38:02] Well, now check this out. [38:04] It's created the PR. And in the instructions, I've told Claude, hey, whenever you create the PR... [38:09] open it in the user's default browser. So now we have our PR opened here. [38:15] And this check to deploy it to Vercel will fail. But that's OK, because I give it one more step here. And all this red looks scary, but it's not. [38:27] I tell Claude to just monitor the CI every 30 seconds or every 60 seconds. [38:32] until all of the checks pass, [38:34] And I tell them the specific checks that I care about. [38:38] And if any of the checks fail, [38:40] Just fix yourself and then push the changes. So you know, if people push something to GitHub and there's a TypeScript error, [38:48] they'll see some error over here in the GitHub UI. They'll take a screenshot, they'll send it to me on Slack, and be like, [38:55] why is my thing not working? I want to just avoid that entirely. [38:59] And, you know, going all the way back to my first principles, like, [39:03] If the AI is asking you to do something, [39:06] like check the PR or tell me the CI status, you should really be thinking about how do I [39:11] teach Claude to just do that for itself. [39:14] So over here, this slash deploy command, [39:17] Literally it's just end to end. I just sit back and watch it loop over and over and over again. [39:23] checking its commit status, its CI status, making sure everything works. And then when all of the check marks over here are green,

[39:32] the script will stop. [39:35] I think this is pretty awesome. I feel, I hope it lowers the barrier and like the intimidation factor of having to learn all these tools, but at the same time, you know, [39:46] If you are curious, you can just sort of read along and understand what's happening. It's like instructed to communicate in clear English what it's doing. [39:54] My favorite part of this, and it's not going to be what people think. I think, [39:58] The slash command is amazing. I think running through all the pre projects. Great. [40:04] I love that you just open it up in a browser window. It's one of those things that, you know, even if you created the branch, created the pull request, [40:12] said it was ready to go. People are like, okay, well, now what do I do? And just forcing open the browser window and saying like, this is where [40:21] It lives on GitHub. My question is, [40:25] Do you have to get your code reviewed? [40:27] in Prototype Playground? Or do you just? For Prototype Playground, no. I mean, people can always ask for it, but no, we pretty much just YOLO merge. I think the thing that I mostly check for is, like, did my PR accidentally mess up someone else's prototype? Yeah. [40:42] But again, like that happened a couple of times and that was annoying. So then we created this clod local file that's like important. Do not do this, you know, and that seems to have fixed the problem. So. [40:56] Yeah, a lot of YOLO enable auto-merge. [41:00] And of course, it's not perfect. I don't know. It seems to be hallucinating some stuff here. Like it thinks...

[41:05] it thinks these passed even though they haven't i don't know it it's close so i'm just gonna zoom out everything that we went over you created a shared repo for your entire team where you could have name level directories no database we're just using [41:20] metadata json and and and shared code to put different prototypes inside this repo [41:27] You've set it up with both global Claude rules as well as local Claude rules plus [41:34] Claude commands and Claude skills to sort of guide people along common paths. My favorite one is going to be Figma. [41:42] to code. It's so beautiful. It's so good. And then the number one rule that I've heard from you today is [41:51] When asked to do something by Claude, [41:54] teach Claude to do it. It's so. So you have this amazing prototype. Um, [42:03] playground, you've set all this stuff up. How has let's let's just do a couple lightning round questions and get you on your way. And my first one is, how is this shift from [42:15] Doing things, you know, maybe exclusively in Figma or in these lower fidelity prototype models. [42:21] to really leaning on things like [42:23] - Claude Code. [42:25] code-based prototyping. [42:27] How has that changed the design? [42:29] team? Has it changed a small part of the design team? Do you feel like overall things in the organization are shifting in a way? How do you feel like it's changing the way people work together?

[42:39] I still use Figma. I probably still spend [42:43] 60 to 70% of my time. [42:46] in Figma [42:47] You know, there's just certain things... [42:50] that you're making... [42:51] that don't need to be in the browser. They don't need to be coded up. You can just look at it and be like, yeah, that's roughly right. We should just ship that. [42:58] I find that [43:01] as you're designing for things that use AI, [43:06] That is not true, though. [43:08] So for example, if you were building [43:10] A... [43:11] chat bot [43:13] or in my case, I work on Notion AI. [43:17] I don't think you can design [43:20] a good chat experience in Figma. [43:23] You can design what the chat input looks like. You could design a little chat bubble and a send button and like a dropdown for model picker. [43:31] I think all that's fine in Figma. But what you can't design in Figma is what it actually will feel like to use that thing. [43:39] I probably should have said this at the very beginning, but the reason Prototype Playground existed is because when I started working on Notion AI, I was literally designing conversations in Figma. You know, it was like... [43:50] the user is going to say this and then the AI is going to say this and then it's going to work perfectly and create a page or a database and like you mock these [43:59] golden paths in [44:01] Figma [44:03] And then the engineer's [44:04] go and they build it, [44:05] And then it just doesn't work that way, right? You send a message, the AI gets stuck, or it asks a follow-up question, or it does the wrong thing and you need to correct it. And...

[44:16] Prototype playground was for me a way to connect to real AI models and just start feeling out like okay [44:22] how are the models going to work if i submit this kind of prompt [44:26] What happens if I connect it to the Notion MCP? It doesn't even know how to create a page. What happens if it runs into an error? Oh, right, we need to design an error state for this. What happens if the model is thinking for two minutes and the user's staring at an empty chat screen? Like, what should we do in that intermediary time to... [44:42] help them feel confident that it's working, that it's doing the right thing. Is there any way to show incremental progress? [44:47] I just found those things [44:49] Very very hard to design in Figma. [44:53] So to go all the way back to answering your question, I think as more and more people are designing [44:58] apps that both [45:00] are for ai or incorporate ai in some way [45:05] They're going to need some other... [45:07] like native [45:09] code first way of working to actually understand what the models can do. [45:14] It feels honestly kind of bad. It feels like a lot of wasted time where... [45:19] every month. [45:20] the whole freaking industry has to learn like, oh, what are the new capabilities of this model 4.3.2 Max Pro? And then a month later, it's all irrelevant because the new thing has come out. And then you learn that. [45:33] Um, [45:34] It feels like a waste of time. Unfortunately, I think it's necessary because [45:38] The model capabilities are still advancing quite steadily with each release, [45:43] And it's really important as designers [45:45] to understand what models are capable of doing so that we can create product experiences and designs that sort of live right at the edge of what the model is going to be able to do well.

[45:56] What's really frustrating is if you design something that's like, [45:59] "Oh, a user's just gonna ask for [46:02] a cool website. [46:04] and it's going to be this perfect output website on the other side. [46:08] Models can't do it, right? Or they require a bunch of fine-tuning and sort of like intermediary prompting to get that right. [46:17] Designers just have to know what's going on under the hood there to [46:21] to design something that's, [46:22] plausible and possible. [46:24] So I suppose... [46:25] the more products incorporate AI, the more designers will have to shift to thinking sort of [46:33] prototype first but probably prototype first with actual code under the hood [46:38] where you can incorporate [46:40] Modern models and see where they break and see where they're good and see where they're bad and actually form an opinion about a [46:45] which models are good for which things, that kind of stuff. [46:49] So speaking of which models are good for which things, and you're using my current fave, [46:54] My babe. [46:55] Opus 4-5. Wine... [46:57] Why, I mean, why Claude code? Why cursor in this non cursory configuration? [47:04] Tell me how you arrived at this is your tool stack. [47:07] I need to play with more of the cursor stuff. I actually think cursor agent mode is pretty awesome. I've like, [47:15] clearly tried it a little bit. I just haven't gotten that, that far. Um, [47:19] The thing that I still really appreciate about Cursor, I actually technically use both. Like if I have, I don't know, like some file

[47:27] And there's... [47:29] there's like some error here. I still really appreciate being able to just hover over the error. And there's a button that says fix and chat. Yeah. That's still faster than like copying and pasting it down into cloud code. So I actually use both a little bit. [47:42] I just think Claude code does the best work. I don't know how else to describe it. There's this weird [47:48] feeling as you use all the different models for different things it's like [47:53] to different people they just feel right for me opus 4 5 is just [47:57] insanely [47:58] good at doing what I want. [48:01] And I like the way it approaches problems. I like the way it plans. I like the way it executes. I like the way it communicates back to me and the follow up questions it asks. [48:11] And then, you know, why not use [48:13] Opus 4.5 [48:15] in cursor versus in the terminal UI. [48:18] I don't know. I think this is just... [48:21] purely personal preference like some people look at this and they're like this looks like shit like give me buttons and ui and components and and drop downs and things like that [48:30] And for me, I don't know, this just feels... [48:32] nice and easy to read. It just feels good. As our friends over at Every say, each model has a mouthfeel. [48:38] - Yeah, yeah, exactly. - Claude Code and Opus have a good one. Okay, and then my very last question, 'cause you seem like a expert prompter. [48:48] But when AI is not listening, when it's not listening, when it makes up, [48:54] you know, CI checks that passed where it didn't actually pass.

[48:58] What is your prompting technique? [49:00] Basically, I notice there's a direct correlation with how good of things I can make and how tired I am. And if I ever get to the point where, man, Claude just sucks, it's doing the wrong stuff, [49:14] And I go back and I reread the thing that I said. I realized I made no sense. And so the best solution for me to write better prompts is like, [49:21] Go to bed. [49:22] Try again tomorrow. [49:25] Which... [49:26] I don't know. [49:27] I don't know if that's a cough out answer. It's not actually writing better prompts, but uh... [49:32] you know, your output's just directly [49:34] correlated with how good of context you give the thing and if you're giving it sleepy tired lazy please fix this type [49:42] commands, it's going to do bad work. [49:44] I don't know if this is what you intended, but you gave me very good both relationship and parenting advice there, which I'm thinking about. I was trying to ask my kid this morning to do something. I'm pretty tired. [49:55] And... [49:56] I clearly... [49:58] the inputs. We're not going to get the outputs that I want. Well, it's easy. I mean, just go take a nap. Can't you do that at any point that you need? I love that. You know, one of my favorite little agents, Devin, does have a [50:11] sleep. You can send the agent to sleep, baby. We just need the agents to send us to sleep. Well, Brian, this has been... [50:18] Awesome. Just a deep dive, I think, in a very forward looking view into how [50:23] design teams, as you say, especially ones that are going to be building AI products are going to start doing their work. So where can we find you and how can we be helpful to you in Notion?

[50:32] You can find me, I'm mostly on... [50:35] Twitter or X, Brian underscore Lovin, or my website, BrianLovin.com. [50:41] And then [50:43] I work on Notion AI, and I think it's genuinely one of the few useful... [50:49] sort of knowledge work agents. So if you haven't tried it, [50:53] try it and send me feedback. We're always trying to make it better, help it do more things better, faster. So try Notion AI, yeah. [51:01] and we're big fans of notion ai too here at the podcast so definitely give it a look and definitely get some feedback [51:07] and we will send it directly to Claude and put it in prototype playground. Brian, thank you for joining. [51:13] How I AI. [51:15] Thank you for having me. Thanks so much for watching. If you enjoyed this show, please like and subscribe here on YouTube, or even better, leave us a comment with your thoughts. [51:24] You can also find this podcast on Apple Podcasts, Spotify, or your favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at howiaipod.com. [51:42] See you next time.