Build Week: What we Made [Part 1] with Dave + Travis

💻 AI side quests, but make it Snuggle. 🤖✨

In this episode, we’re fresh off the Snuggle (our company retreat) and talking about what happens when you give a curious team some time, space… and a bunch of AI tools. Travis and Dave join us to share how experimenting with tools like G-Stack, Spec-Kit, and Claude helped us explore ideas, prototype faster, and take on projects we normally wouldn’t have time for.

We get into the practical side of it all—what worked, what didn’t, and where AI actually fits into a creative workflow (without the hype). It’s part experiment recap, part philosophy, and part “what if we just tried it?”

Whether you're building products, designing systems, or just curious how AI can fit into your process, this one’s for you.

🎙️ What we cover in this episode:

• 🧑‍🚀 Why we dedicated Snuggle time to AI exploration
• 🧩 How tools like G-Stack, Spec-Kit, and Claude fit into real workflows
• 💻 Using AI to prototype and pressure-test ideas quickly
• ⚖️ The balance between human creativity and machine assistance
• 🧠 What surprised us (and what didn’t)

⏱️ Timestamps:
00:00:00 – Welcome & Intro
00:01:59 – What are Snacktivites?
00:02:57 – Company Behaviors Framework
00:04:16 – Build Week Explained
00:05:00 – Travis's Thoughts on AI Shift
00:08:05 – Ethics Over Slop
00:08:31 – Education and Intent
00:11:47 – Using AI as a Force Multiplier
00:14:50 – Skills Files and Standards
00:21:10 – Tooling Roundup: G-Stack, Spec-Kit, and more
00:25:46 – Color Awesome Finale
00:27:45 – Wrap Up ... To Be Continued in Part 2

🔗 Links & resources:
Font Awesome: https://fontawesome.com/
Browse icons: https://fontawesome.com/icons
Kits: https://fontawesome.com/kits
Web Awesome: https://webawesome.com/

Credits:
🎵 Theme music by Ronnie Martin
🎶 Interstitials by Zach Malm

Stay up to date on all the Font Awesomeness! 

Transcript

Dave Gandy: [00:00:00] think the same is true for AI as it goes forward. What are we doing with this and why? The intent behind this, what we're trying to do, um, is I think gonna help guide a lot of what that usage look like-- looks like and what even, you know, kind of when it's a good choice and when it's not.

Matt: I've got Dave and Travis on Podcast Awesome at the same time, which

Dave Gandy: no way.

Matt: nearly a miracle. So

Travis Chase: And

Matt: for coming on the podcast. 

We, uh, we're just coming off our last springtime snuggle in Bentonville, where we all meet up in person, and it's always an awesome time. And, um, for folks that are not following along w-uh, with the

Dave Gandy: Yeah, so, uh, the 

Yeah, so the reason we call them snacktivities is because they are little bite-size, [00:02:00] uh, projects small enough to just be able to get done in a week. Uh, so a lot of times they involve kind of collaboration between a couple of different people, um, because it's kinda neat whenever you're able to, you-- Everybody's in person, there are things that you can only uniquely do when you're in person, so we try to take advantage of that, and that's what snacktivities have typically been.

But when everybody's in, there's a lot of stuff going on, and it's generally pretty hard to get that done. So usually there's maybe one or two that people finish in a given week. Uh, they turn out pretty neat. We've got some really cool features that have come out of those. Um, the, uh, Font Awesome Reusable Tidbits and, uh, Parts.

Font Awesome Reusable Tidbits, uh, other one, o-otherwise known as Icon Farts. Wait, I'm... Nope, it got renamed. It's the Icon Wizard. Um,

 so, uh, last week was really cool, well, there's sort of a, a thing that's been happening in our industry, and we've been kind of curious about it. 

Um, so we have, uh, there are three behaviors that we have at the company that we sort of expect everyone to embody, right? This is something that we want to practice.

It changes how we work together, it changes how we design, how we, uh, develop. It changes kind of all, all, all pieces of this. And so those three behaviors, Travis, give me one of them.

Travis Chase: want me to keep going?

Dave Gandy: Humble and helpful. You got one, Matt? No, I want Matt to do one.

Travis Chase: Ah.

Matt: humble and helpful, curious, and curious and reasonable.

Dave Gandy: Yep. And the last one is adventurous and dependable. So nice job team. Um, so they're really important. These are paired together because any one word doesn't quite get what we're going for. Um, curious and reasonable is that before jumping to a conclusion on a thing, we're gonna investigate. We're gonna be very reasonable.

We're not gonna jump to sort of a dogmatic perspective without really spending time and getting some experience with it, because it's It's very easy to have opinions about something that you don't know that much about. And so we wanna be curious, and we wanna find out and make sure that, you know, we've got some opinions that are pretty well informed.[00:04:00]

Uh, adventurous and dependable, right? Um, uh, we want to, uh, try new things, go new places, try out something we haven't done before, but not in sort of a crazy way. Uh, but we wanna do it in a dependable way, right? 

we called this adventure, uh, we called it Build Week. So everybody came in, uh, and we had, uh, you know, a four or five-hour session a day where people could just work and build.

Uh, and then at the end, uh, we did show and tell and showed off the projects people worked on. Um, so yeah, it was pretty neat.

Matt: Yeah.

Travis Chase: the really interesting thing about it was we wanna make sure, uh, in the spirit of exploration and the spirit of like, just you, you don't have to worry about working on something for Font Awesome, Web Awesome, or Build Awesome. We-- We just like, you could work on anything. So that way it's just super easy, and there's no like, uh, you know, importance necessarily set to it other than learning and checking out these tools.

Matt: Yeah. Travis, I seem to remember like In the last year or so, um, as the-- Travis is known as the reasonable, the reasonable person on the team and is like really, um, has a real strength and kinda gift for thinking clearly about things.

And remember you saying at one of the snuggles, um, it was a year, year and a half ago, you were saying, "You know, everybody knows our industry is changing, "It's not really the ideal, and it's not necessarily what I would prefer, but it is the reality, and let's just stay curious about this and see where it goes, and let's learn." can you reflect a little bit on your thinking has been like in regards to just AI in general and how you've approached it and how you've tried to set the tone for it?

Travis Chase: Yeah, absolutely. Uh, I think it's interesting that, you know, it has come such a long way and, uh, one thing about being reasonable is also having, uh, reserve the right to change your opinion. It probably still, and I've gone on record as saying it may not [00:06:00] necessarily be the world, uh, in which I would have envisioned or I would have picked.

Um, you know, it's, it's, it, it's weird on how it, it's rolled out. Like, there's some really great things about it, but there-- you're definitely, you gotta acknowledge the, the crazy things about it as far as, you know, jobs, what it means for developers, what it means for designers, and the things that we used to really like to do, or at least we thought we liked to do. Uh, you know, now we have agents that do them for us and, and we're doing more product management and product guidance, which actually can also be really fun. And so they've just gotten better and better and better, and I wanted to make sure, and especially with talking with Dave as we started even playing with these tools more ourselves, it's like, let's just, let's just have people build with them.

Let's, let's, let's see what people discover. Let's see what people find. Let's remove maybe some stigmas or some, or the things that are scary or, and just see like how helpful this thing can be that maybe we can accomplish more. Because one thing we've always had the problem with, uh, especially with a small company, most people that run small companies know this, you have an infinite backlog, and there's things that you've wanted to get to for years and years and years, but they

Dave Gandy: Decade

Travis Chase: Yeah. It just, it, it, it always happens. And so what's exciting about this now when I look at it is It may allow us to, uh, get more of those things that we wanna get done from our backlog that we've always wanted to get done, and really get to expand, uh, the offerings not only for our customers, but, uh, things that we get to work on and even different exploration areas where now that you have a really good team with you, you can maybe explore a new language that you were maybe too scared to like dive in and learn.

But now it's like, well, you can get the help and you can get it, it, the coding, and you can get the, uh, you know, the guidance that you need and feel like you can still ship really good product. Um, you know, and but still have, you know, you still have the brakes, you still have control, you still have review, you still control the PR, you still say what goes into production. Um, but you can get there maybe a little bit more fuller, a little [00:08:00] bit qu-- a-and probably a lot quicker.

Matt: Yeah. 

Folks will trust companies that have, like, an ethical mindset about how to use these tools.

They're like, "Oh, they're, they're using that as a, as a tool. They're not using it as a replacement and just creating a bunch of slop," right? 'Cause clearly there

Dave Gandy: Yeah.

Matt: AI garbage, and it sorta kinda works, but it has-- there's no personality, there's no soul to it.

Dave Gandy: Yeah.

Matt: you guys have, like, any thoughts on that?

Dave Gandy: Yeah, I think, um, I mean, the sa- the same is true if you start seeing AI usage in like high school, right? your English class, right? If the goal of your English class is to write papers, as someone writing those papers, you can use AI, where it writes that paper for you, and then you turn it in. And then you can get into an arms race where you try to detect and not, and then you put it through another pass to make it not be detectable by A, what- whatever, right?

It's an arms race at that point, and it's also a waste of time. Um, if you instead use it a different way, you can use it to take what you've already written and make specific suggestions to you, right, with prompting. Ask it to, uh, review according, you know, to some different things and help it to improve.

Since your goal here is learning how to write, you don't wanna replace that thing, right? So you gotta think about what the goal of these things are. And, and so, um, education's gonna have to change, right? You're gonna need to write a paper in class. Um, that's the only way to know that it was done there, and that's also, like, a great way to do it, right?

Little, small, bite-size. How clear can you think in an hour? Go. Put it on paper. Don't type it. Write it with your, your grimy little hands from, like, centuries ago that were good back then and still are now.

Matt: Right.

Dave Gandy: think the same is true for AI as it goes forward. What are we doing with this and why? The intent behind this, what we're trying to do, um, is I think gonna help guide a lot of what that usage look like-- looks like and what even, you know, kind of when it's a good choice and when it's not.

I- if the point of, of what we're trying to do [00:10:00] is learn how to code, uh, we probably don't wanna rely on AI so much out of the gate, right? Um, but if the point of it is to build software, that's a different thing. That's a very, very different aim, right? If we are focused on our customers, and if we're focused on building things for people, um, then how long it took us to write it is not really relevant to the end user, right?

As a matter of fact, it is relevant in that we can maybe ship two, five, ten, a hundred times what we were doing before. That's within spitting distance now in a really kind of funny, strange way. So we've gotta decide who are we doing this for? Is this job about us and how we enjoy working and what's fun for us, right?

Is that what most jobs usually are? Um, or is it about what we can do for our customers, the problems we can solve, and how much more we can help them? Um, and I think that, for me, helps... has helped me think through some of the... I can't do a lot in life without a framework of what right and wrong actually looks like.

Um, and that's been sort of my journey through this, is who am I focused on? Where am I focused? As an engineer, right, the best engineers in the world, the number one, uh, quality is humility It's not to build the way I think is fun, but it's gonna build it to be to build in a way that is most useful for somebody else.

And so far, building with this, um, uh, you know, I, I had, I had a very, very interesting time last week. Uh, and there's a, you know, there's a lot to talk about from what, you know, kind of I experienced last week even personally. But Travis, what do you think?

Travis Chase: Yeah, I think I look at it as being a, a force multiplier. you know, and it, it does necessarily make us think about how we change, how we look at hiring, how we look at team construction, [00:12:00] even how we're looking at project

Dave Gandy: Yeah.

Travis Chase: It's, it's

Dave Gandy: Yeah,

Travis Chase: making us think about how we might do things a little bit differently than we've done in the past. And it-- I, I think for me, like a lot of technology, it's, it's a tool to accomplish a goal and, right? Our, our goal is to be the de facto standard, uh, the de facto toolset for web builders. And so if we wanna do that, then we need to get those tools in their hands, and we need to make them simple, we need to make them powerful, and we need to save people time and them accomplishing their goals, which their goals is to build something for other people, you know, on the web.

And so using these tools as a force multiplier and getting that out there, um, I think is a win for us.

Dave Gandy: I think the interesting thing, it's, it's really, really good that Travis mentioned that. I think the thing I found most interesting last week with all the projects were, uh, whatever somebody was, right, whatever their, whatever they focus on, whether it's design, whether it's development, whether it's security, whether it's automation, ops, whatever the thing is that they know best and, and, um, right, have spent the most time with, that's mostly where they spent the most time with AI, right?

Most of the time that they spent, uh, instructing it what they knew well, but then they were able to get so much more done because of all of the other pieces that may, may not be their areas of expertise, um, they could sort of take some, uh, take, take, take the lead from AI for. But, you know, last week I spent all of my time on user experience.

What's the interface behind this thing I've been wanting to build for ages? You know, we talked even for a couple of years of hiring somebody to focus on this one particular product, and I got a version of it knocked out last week, uh, fast enough in a way that was, um, well, there's more to be done now.

But, uh, there's this part in the design process where you get enough done that you jump to the next level. "Oh, now I need this. Now I wanna see this. Now I wanna investigate this. Does this work? How well is this working so far?" Uh, and it just speeds [00:14:00] up that iteration process so much, right? And it opens up the flexibility of these things, um- One of the biggest problems with old school software development is how rigid it has to be, right?

Well, we've got to start with design. Design has to go through and make some screens in Figma, and they've got to be pixel perfect, and then that pixel perfection must then be replicated by dev in every single way, in every single, you know, every single down to every pixel. And why didn't you do it that way?

Well, Figma doesn't work the way the web does. Well, that shouldn't matter. You should get it done anyway. Um, and this sort of like a-adversarial relationship between all of the workers, uh, doesn't need to exist anymore. There are steps that we can skip. We can jump straight down and do this all out of order, and people come in in different places.

It frees us up in a way that was-- that I find fascinating.

Matt: Yeah. It makes me think about like in a broader sense that as new tools are, are made available to folks, you know, I think of like, um, uh, maybe it's a little bit of a stretch, but like web components as an example. You, you start to get into these conversations of, um Kind of gatekeeping people that are sort of grumpy and curmudgeonly

Dave Gandy: Yeah. Yeah.

Matt: you need to use this, uh, tech stack and do it properly."

And if the goal-- I-- It seems like a pretty good goal if you can kind of flatten things and democratize things and say, there are people out there that want to create cool stuff on the web. There are people that also want to create not cool stuff and be creepy. But like, for the most part, if, if we put a positive spin on it, people wanna create stuff and they wanna share.

They wanna make the web a better

Dave Gandy: Right.

Matt: So if you can create tools to lower the bar and, uh, get everybody creating, I don't know, it makes me think of like, you know, if you're in art class, uh, and, and the, the teacher is like... And you're working with [00:16:00] kids, and you just want them to be creative and create, but the art teacher is, like, so particular about theory and stuff that it's just a bummer for everybody and nobody has any fun.

You don't create anything. Uh, it, it's sort of like this idea of, like, if you can get tools in the hands of people to lower that bar, so more people can create, um, that seems like a good thing. Now, let's do this in an ethical way and continue to have conversations, but, um, it, it just-- It like, uh... We, we gotta at least, like, be thinking about the positive aspect of

Dave Gandy: So

Matt: and

Dave Gandy: interesting thing here is, the interesting thing here is that problem, right? Having tho-those folks at the organization that on the outside look like these old curmudgeons that are just always up-upfront, upset and frustrated. Actually, they can be happier than ever now because they write a skills file For the technology you wanna use, and then AI will just use it, right?

You set your security standards, your design standards, your, your... Right? So I didn't know what I wanted. Travis was like, "No, uh, do Rails. Let's use Turbo, and we'll use De- Kamal for deployment." Oh, okay. I don't know any of that personally, right? But if you wanna talk about UX, I mean, I, I wanna-- let's, gonna go, uh, Nielsen Norman Group, usability, don't make me think, right?

There's a hierarchy of things we wanna go down for usability and, you know, get all that right too, right? Um, now we can just have all those things, right? You can have the curmudgeons, you can just have them, right? Uh, you can have the curmudgeons write their skills file, right? You can have the designers write their skills file.

You can write your brand file, right? Your design.md, right? You can write all these specs up for these things, and then it'll just do them, and it won't complain. It's actually kinda neat, right? Like, you spend the time with your organization developing these, finding the right place, um, and then, uh, you get this force multiplier even more on that expertise, not less.

That's the thing I've sort of been blown away is how, how it helps you be more of you, [00:18:00] not less. It seems like, oh, AI's just gonna go and replace what you do. No, it doesn't work that way. It actually helps you become more of what you are. Every security person, better security, right? Doing, doing the best with security, and now we can replicate it in the organization.

Uh, on the development side, best design, how we wanna do this, they, they write that, we replicate it across the organization. User experience, DevOps, right? All of these things, each one of those people can now be more of that. It encourages us to document these things and spread them in a very specific way.

Uh, that's kind of exciting that we could just have all these things now.

Um, yeah, I, I don't, I don't think I've ever been this excited to build ever before. You'd think, "Oh, it's no fun. It's doing all the work for you." It doesn't work that way. I, I hate to tell you, right? Like, it just doesn't work that way.

 kinda neat how, uh, yeah, it actually helps you be more of who you are in, in my experience.

Matt: Mm-hmm. Wow!

Travis Chase: mean, early on I would ship little prototypes, uh, to prove out a concept, put it in front of

Matt: sounds [00:20:00] great. 

what tools were folks using and what were they building? Maybe we could

Dave Gandy: I think there were three primary tools, I think. So there was, um, the one I spent all of the time with was G Stack, which is open source created by Gary Tan leveraging Claude. Uh, Travis, there was Spec Kit. You wanna talk about that one?

Travis Chase: Kit with, um, e-especially for some like, uh, UI type dev with, uh, Pencil.dev, which is a tool or S- or Skitch. also just kinda like for those that haven't spent a lot of time with it and just w- you know, were starting their journey, it's just kinda fire hose in, in, you know, Claude Code itself or in Claude Desktop, where you're just-- you're tell-- you're seeing what happens, and then you realize real quickly like, "Hmm, to get better results, maybe if I plan mode."

Well then if from plan mode, oh, Spec Kit is like plan mode on steroids a little bit, you know. And then G Stack is like, okay, well, now we're thinking about like whole [00:22:00] product releases, and it's just, uh, the tooling

Dave Gandy: And business model, go to market

Travis Chase: Yeah. And how you mix and match them, um, and how you-- how they wind up.

'Cause I, I personally have really had a lot of fun, uh, recently. I, I've done a couple different projects, um, one using Pencils and Spec Kit, and recently I've been really messing around with G Stack and tied in with, uh, Superpowers, uh, which is like this really great, uh, planning tool when you start to get into the implementation phase.

And so, like mixing those two together to build whole, uh, platforms is just been really, really fun. Uh, 'cause after decades of coding, um, you know, kinda up to the point where it's like, "Uh, maybe I don't need to write that form for the one millionth time." I'll kinda let it write the form, but I'll tell it what I want to collect and how I want it to behave and how-- what I want the user to flow and feel as they go through this thing, and what I want to ultimately accomplish, and I'll let it do, uh, the form in whatever chosen platform.

Uh, recently I've been messing around with, uh, just doing stuff in Rails, um, just, just to see how it like manages, uh, that. And I've done things with Elixir, and I've done things with JavaScript, all that kind of stuff.

Matt: Yeah. How about you,

Dave Gandy: Yeah.

Matt: what tools were you using and what did, what, what were you working on?

Dave Gandy: Yeah. Yeah, so I exclusively worked with G Stack. Um, it's also, like, the way I think. It's, it's sort of like what I was saying before, like, I have a hard time operating in the world unless I've already got a framework for what right and wrong looks like, what wisdom, right? There's the, there's the black and white right and wrong, and then there's, like, wisdom.

If I don't have a framework I can work from, I have a very hard time making choices until I get that sorted out. But once I get that sorted out, I can move, I can move quickly, and I can move what I feel like is with confidence. And so, um, G Stack kind of works that way, right? It's asking the right high-level questions to make sure, are you working on the right thing?

Uh, there's a great Gary Tan essay called "Boil the Ocean," right? Uh, which is, you know, it used to be with y- needing to ration our resources for development, we need to take a teaspoon of water at a time and then boil it and put that down and go get [00:24:00] another one. And now with the speed at which we can work, it's sort of like, no, no, no, start with the biggest thing possible.

Let's boil the ocean. What does that look like? Um, super interesting essay. And, um, and the reality is, you know, a little more challenging than that certainly than just saying it. Um, uh, but it was really neat to have a lot of these questions. Uh, and it's als- it also takes you through the same process that Y Combinator uses, that they've been replicating now as it's a trillion dollars of net worth in the companies over the last, you know, 15 years.

It's, it's a stupid uncountable number. Um, and, uh, so there's some experience there. Whatever you believe about individual specific people, especially their character, right? Is there an aspect of who they are that you can still learn from? No matter what you think of their, uh, uh, of what you know of them, of your maybe snap judgment about not having, uh, you know, all of the information about them as humans, 'cause we don't really ever get that, um- What does that actually look like, right?

And, and so it, uh, G Stack is, is, I think is, is super fascinating. Um, it already has this set of built-in skill files that, uh, Gary's been working on that are outstanding. I've produced the best... So, uh, my focus is primarily user experience, not UI design. Um, and so, uh, what was neat is, like, it, it made me a better UI designer, right?

The stuff that I was able to produce with, uh, with even some very specific instructions, the rest of what it did around it was better than I was gonna do on my own. Um, but I got to get really specific really quick with all of the micro-interactions here to the point it's like, it's working, it's jamming, this is cool.

Um, there's a tool I've been wanting to make for a while called Color Awesome. Um, and this is for helping people get, uh, a really great color palette for their website. Um, and that's the, you know, that's the, that's the right brain. The, the, the, the left brain. Uh, wait, which is which? Which is left brain and right brain?

I can [00:26:00] never remember. Okay, so, so that's the right brain side. Um, the, it, that's the creative side of the brain, right? The what, what, what looks good. But then you've got the, the, the other side of the brain, the technical, yeah, but what's gonna work? That's a really nice, you know, shade of yellow you put on white, it's not readable.

Humans can't read that. No one can read that.

Matt: Yeah.

Dave Gandy: accessibility rules automatically underneath this so that what you just made now is gonna work for everybody in an accessible way, and we just handle it. You don't even notice it, right? That's the thing I've been wanting to work on for a while.

And I spent the week building that, um, and then ended up with some stuff that, like, felt really good. Some of the previews I made of these color palettes were different than what I've seen from anywhere else, right? I wasn't taking other people's designs. I was, um... One of the things I, you know, I teach the, the, you know, the, the first Lego League team is we wanna stay relentlessly focused on function first.

What does this need to do? And watch. Watch what happens. When we do this, form falls out of it for free. You get a form that falls out of it that's different than what anybody's seen, but it's also so cool because it's so related to how it works that it's exciting. And I got to do that last week, and I was so excited.

And what it ended up, uh, what I ended up making ended up looking like stained glass, right? These, um, these color palettes, as you put into there, literally looked like stained glass, the visual, the visualization of them, which was kinda neat because I just went to the Sagrada Família a couple weeks ago, and that's where, like, my brain and mind and heart had been.

And it was neat to see what just kind of naturally fell out of this was the combination of all these things was me, but more me. That's the bigger shock of all of this for me personally, I think.