Ep. #6, The Weird Edges of Taste with Betty Junod

Betty Junod: Hi, I'm Betty Junod. Thanks for having me, you guys. I would say I'm tech curious, tech optimist, tech enthusiast. I've been in marketing, helping talk about what it is, these magical things that people are building and trying to make it understandable for regular humans.

Adam Zimman: Absolutely. And something that caught our eye and we wanted to specifically talk with you about, or at least a point of starting, was you wrote a blog post about what happens in this world of agentic code generation when you start to think about your applications and the users or user base of those applications.

You know, in the marketing world, talk about ideal customer profile. And you raise this point of, well, what happens if your ideal customer profile is one person? And that was something that was really intriguing to us.

So why don't you tell us and give those that haven't had a chance to read the blog post, we will put a link in the show notes, a little bit of a brief recap and also what you were thinking about as you were coming up with that idea.

Betty: It's like my journey in exploring all this new stuff. And like I said, I'm a tech enthusiast, but I am in marketing and I'm one of those marketing people that have always been in marketing, never coded before, never did any of that stuff, but instead I always sat next to an engineer or product manager or what have you, and just said like, "what is this thing you're building? Show me."

Maybe I could get to a place where I could demo the thing myself, depending on what the product is. But the idea of building software was something that was always just, it's like jumping over the canyon, right? And there's this other side of the canyon where-- What's a terminal? CLI? I've seen it. I know what these are conceptually. I know what they are physically. But I never learned that language.

So it's learning a foreign language that I never learned. And it just happens that those words build something that people can touch and interact with. That was always a gap. I had lots of ideas. I was also very frustrated with some of the software tools that I had maybe acquired either personally or through work for my job.

And the most interesting thing is like, oh, with what we have with agentic coding is like, I can go to one of these tools and say, "hey, I want this, and just write it out," and it gives me something. And then I keep playing around with it.

So the idea that I can serve myself, and the barrier to entry is not, hey you're going to have to spend all this time to learn the whole language and then understand architecturally what that means. Like, what ports do you need to open or not, or this or blah, blah, blah. Who cares? If it's just for me, who cares? It's just a matter of what my credit card can handle.

And I built an app. I think I had four people use it, and it's like 19 cents for the day.

Adam: Reasonable cost.

Betty: Yeah.

I'm not building this to sell for other people, which is where a lot of other questions and concerns come up. I'm not trying to run a business. I'm just trying to answer my own question.

Adam: Well, I mean, at the end of the day, I think one could argue that the vast majority of software is being built to enable people to do a task, do a thing. Whether that's for entertainment purposes or business value doesn't really matter. They're just trying to accomplish something. They're trying to get something done.

And so the question is, if you were able to build this thing that accomplished the task that you had at hand, I would argue that that's of just as much, if not greater value to you than going and procuring something from another company.

Betty: No. 100%.

Heidi Waterhouse: Okay. But what's the app do?

Betty: Ah. So I can't remember which app. I may have put in a little like job search app in there. But one of the things I did build recently is, as a marketing leader, the hardest question to answer is like, what's our current share of voice and sentiment? And it's really hard because it's hard to do. And it's something that you maybe get a once a quarter report from your PR firm because it's hard to do.

And it's also something that is like there's tools out there that track social sentiment, what have you. I've just never been-- And a lot of those tools go in great depth. But it's like sometimes like your executive leadership doesn't want to go through all the depths. They want a snapshot and they want to know like, "oh my God, somebody posted something, now it's going viral. What are we doing about it?"

Do you know what I mean? They just want snapshots so that they know that everybody has tabs on the business and are reacting to real time information. So the thing I built was something called-- The app actually named itself Brand Pulse. I'm like, "not bad," but it's a three tab dashboard, effectively.

One is the current state of the company or whatever, like you know, the coverage, red, yellow, green on that coverage and then pulling up a set of articles or things, when it's covered, that are the most engaged. Because I don't care about all the coverage, I care about what's the ones that people are picking up. Right?

And most engaged would mean likes, shares and comments. And then like a little bit of like risk or opportunity because it'll scan the comments and say like, "hey, hmm, the comments are actually like trending bad," or "these seem good, but it's being well received."

The second tab is really just a competitive snapshot. Who are my competitors? Who do I think are my competitors? And then say, "what action is happening with those companies? And then give me like a sentiment on those announcements effectively," again, looking at by how engaged they are.

If my competitor did a bunch of announcements, but literally no one's picked it up, there's been no comments, then it'd say it's just a yellow, like don't even worry about it. It will say like, hey, your competitor announced something, it's getting a lot of traction. Here's the thing that said, here's what is a potential risk or opportunity and an idea for how to address.

And then the last tab was just: What's the conversation? So scanning social media sites, Reddit, Stack, BlueSky, LinkedIn, and based on a set of keywords that I provided, what's the conversation happening with those?

Heidi: That seems incredibly useful.

Adam: Yeah, I like how Betty refers to this as something that's like, only useful to her.

Heidi: Ha. Yeah, as a fellow marketer, I'm like, hello.

Kim Harrison: Can we test it?

Betty: I'll give you the link so y' all can play with it. The thing that I think about, and especially with the book y' all wrote is, the minute I gave it to somebody else: What if it also did this? And I'll be like, get out of here!

Adam: Haha!

Betty: Even with your subscription license, your subscription costs, and then your feature management, that is not what this is about. I am happy to share with you the app prompts. You can run it yourselves inside your company. If you're using whatever tools you all are using, then you can edit it to your heart's content.

Adam: I mean, I think the thing that's really fascinating to me is that you have built this for the use case that you have and for something that is obviously very much in your domain, but the general idea of it and the general premise of it, I've heard been in numerous conversations with people that are building actual products that are doing exactly the same underlying architecture or model to be able to do analysis and feedback for product development.

Like thinking about what features are next or whether or not new features that were released were being used. Yes, it has marketing value obviously as well, but thinking about it from a product and development side in terms of like, where should we be prioritizing our next investment? It's the same questions. And ultimately I think that a lot of times it's the same data sources even.

Heidi: Mhm. And I'm thinking about it like, we can't figure out how well our book is selling. I wonder how I could tweak this to figure out how well the book is doing? Is anybody talking about it? Is anybody reading it?

Betty: Yeah.

Heidi: But all of these are not the thing you wrote. And I think the ideal customer profile of one idea is super interesting because we're saying what is useful just to me, and what if I make a thing that isn't a product, which is so antithetical to the way we've been thinking for 20 years that I'm excited about it.

Betty: Yeah.

Adam: Well, it's not only what if it's not a product, but it starts to get into that weird fuzzy space of , "I just asked the robot to do a thing for me. And I don't actually care how it does it. I just want to have some confidence in the response, that it's generating an outcome that is reasonable in terms of trust and accuracy and then is useful to me for answering the questions or doing the task that I have at hand."

Betty: 100%. And you know what's funny is I made this app. I've been playing around with what this app could look like. I've tried the same app prompts on multiple different platforms to see how different the outputs were. And it was funny to catch some of them in lies. Like, "hey, this looks like fake information." They're like, "you're right. You caught me in a lie."

Adam: Haha!

Betty: I'm like, "but the UI looks nice. Thanks." But this was like, February. February, I was making this into an app. Cause I was like, okay, that's what you do. You vibe code an app. Now I think, "hey, you know what? This could actually be an agent and a set of skills."

So to your point about, it's like an assistant doing some things, right? Doing the research for you and giving you a summary. And like, the thing is now you're just going to get a notification on some, like, "here's some hotspots to investigate. Maybe these can be topics for the, I don't know, the next meeting." Like a team check-in you have, so that everyone's on the same page with something.

But now the thinking is different. Maybe it's not an app. Maybe these are just some utilities. And what I have is: Instead of these specialized apps I buy as a person or whatever, I have a platform or some generic place where I can go make whatever little utilities I need.

James Governor: Amazing. What does this have to do with Progressive Delivery?That's my big question.

Betty: In some senses I'm in complete control of what's delivered to me. So, you know the stuff in your book about giving people the right thing at the right time. When you're thinking about a company building a product, they are spending a lot of time trying to rationalize an ICP as a: Where is the overlapping Venn diagram? If we say our product is for IT operators, that's a broad landscape, right?

And then you say, then from that you have to say like the overlapping place is the ICP and problem area we're going to solve and we're going to build a product for that. And then it's like, do they want all the features that you're making for them when you're shipping them?

This is super interesting because it's now it's like I'm going to build the exact thing I need when I need it. I control when I get it delivered and consumed. I am pretty regularly futzing with my app, as an example, and I'm trying different ways that it can behave. Like I said, should it be an app dashboard, should it be an agent, should it be skills, I don't know, tasks that run?

I'm playing with that as the underlying technology changes, enabling technology changes. But I control when it's delivered to me. I think that makes it interesting from a progressive standpoint.

Adam: Yeah, I mean you think about it like we're talking about the whole name of this podcast is Third Loop, thinking about user adoption. And when you are the only user and you are wanting something new and you get to control what that is and when you get it or when you want it, I think this is the extreme example of that notion of paying attention to the third loop.

You're owning it. Everything you're doing is about the needs that you have and the outcomes that you're trying to create.

James: Betty, you said you tried different platforms. Where did you land on as a self confessed non-software developer for the platform that would enable you to build and maintain and extend an application? You know, who's getting your few dollars?

Betty: Like everybody else, I'm giving a few dollars to a lot of places right now because nothing is free, truly, if you want to have any control of your the settings on what you can do with the information and your usage. But besides that, the app itself is running on Google AI Studio. When I first built this I was vibing with AI Studio, Claude, and Kiro and just each one of those are built for a slightly different knowledge base of the person using it.

You know Kiro is very much a developer IDE. You can vibe within it. But like silly thing, I was like what's a home directory? Because it was telling me to do these things and I was like-- I give it prompts but it wanted me to do some things that were very much like "install Node."

I'm like, at least I know what that is. But imagine if it was like, I don't know, like "your mom," you know what I mean? Because she heard you can do some stuff with like this little chat thing, this AI chat thing. Unless your mom was in engineering, they're going to be like, what's node? Or if you're not an engineering at all, what is node? Right?

It wanted all these like CLI tools. Claude wanted less of that. But at a first pass it made beautiful UI. And some of the things, I'm like, hey, I think this is fake information.

Adam: Yeah.

Betty: And then the other thing was like with Google, like it had the "build it" and then I can preview it. Claude also had a preview mode in the desktop. And then with Google I could just publish and then it's live. Could you imagine if I then had to figure out or actually go through the process of like, "let me get some infrastructure, let me configure my stuff so that I don't--"

Heidi: "Where's my Netlify?"

Adam: Yeah.

Betty: Let me mess up like securing access to my servers. Like, good God, that's asking for, yeah, a whole other set of problems.

Kim: Does it explain any of that or does it just give you like, "here you go, go play." Or does it try to teach you what it did?

Betty: I have asked some of the tools like how do I do these things and that thing. Google was the most seamless with their AI studio that it is meant for non developers to go there. Let me put it that way. The other ones do give explanation but they're still very engineering developer centric.

The idea to me is like, so again this is company building something broadcasting out to the universe. Who are you really building for? I think when we say vibe coding, the idea that it's still coding means the thinking is still that it's developer constructs.

Like why do I need to figure out like which directory packages this, that? Thank you for telling me, that you're picking the making these choices. I don't know what these are, nor do I care.

I will say once I published on Google AI Studio, I then got access to another little icon that I clicked on and I said, oh dear God, I'm now in like some GCP management dashboard and I want none of that except for the billing to see how much this app is costing me.

Heidi: I wonder if you can turn that down. Like shush.

Adam: Haha!

Betty: Well, I have to click on it. So, I mean, from a progressive standpoint, it's there but it's not forced upon me. And then during the deploy they did say like "put in your credit card" of course, because now I'm going to start consuming compute. And then it gave me some options to pick like cost guard rails.

James: Oh it did.

Kim: Nice.

Betty: Yeah.

Adam: So by default it's like don't max out my credit card. You can actually have that be a thing.

Betty: I did like a burner cloud account. It's like a prepaid amount and once that's done, that's it. But yeah, you can set limits on like no more than this much a day or whatever. So from the perspective of like, who is she? Who is this new universe for? Yeah.

James: How much time do you spend messing around with your app? Like is it a bit addictive? Are you saying to your spouse, your kid, "sorry, I just gotta, gotta fix my app." How addictive is this being in control of the thing? How much time are you spending with your application?

Betty: So I will block out time for it within the day just to do this stuff because with the way that my attention works, if it's not in my calendar, then it's just real free range and that could be problematic. So I make time like, hey, I will do research and I will play within these two hour chunks.

It can very quickly suck you in. You know how people talk about how addicting doom scrolling is and social media, like a social algorithm? It's like that but instead of just watching, it is chatting with you.

Adam: Mhm.

Betty: And it's like this always a positive feedback loop. And so in that sense like you can get going. And the other day I meant to just make one edit or run a task on something to help with some research. And then the initial plan was like, oh, I don't know. So I'm like here, let's do this other thing.

I got stuck in like, I don't know, it was like an hour and a half of like basically like "no, that's still not right. That's still not right. I'm super frustrated. But that's interesting that you, the thing you did give me is interesting, but that's still not what I wanted."

And so then it's like, I'm like, we can stop this now thing. Like you're both fascinated and super frustrated sometimes. And then there's other times where two sentences and you're like whoa.

Adam: Yeah, it's like talking to your favorite ADHD friend. You know, like we wanted to know what they wanted for lunch. And then all of a sudden you feel like you just like received the complete history on submarines. Haha!

Betty: Yes. And you know, I'm guilty of like bouncing topics. So it's like talking to myself all day.

James: Adam, do you have friends that are not ADHD?

Adam: I mean, I prefer to them as acquaintances. Haha!

Heidi: Haha!

James: You know, I'm in this industry a fair bit and I mean, good luck with finding people that don't have some issues with their attention.

Adam: Haha. Yeah.

Kim: We're enthusiastic. Okay?

Adam: Exactly.

Heidi: All right, I have a follow up question for Betty. How much time do you spend playing with the tool versus reading the outcomes of the tool?

Betty: Reading the outcomes meaning like what the--

Heidi: The thing that you're asking it to research, like how do you feel that reward ratio is?

Betty: The input is pretty-- It's, for a usable thing, I'd say like 75% to like 85% thing, the lift is low. Very low in my opinion. I'm using it right now for a fair amount of like research as well as like, I use the tool that I, the app that I'm talking about, I use that app quite a bit.

Someone tells me about a company I've never heard about, beep boop beep boop. you know? And I go through and click through all the things.

Heidi: So yeah, you're getting more value out of it than you're spending time on it.

Betty: Yeah. Where I have been frustrated is really, the stuff is moving so fast. And so it's like learning the new thing. So I take something I already knew or something I already built and see like, how would it work in this new construct? And that's where I'm getting super frustrated, because it's new and, gosh an ode to Progressive Delivery-- If you're using Claude desktop, you have to relaunch that thing multiple times a day.

Heidi: Geez.

Betty: That's how much they're releasing. Right? So you may have like two days of no releases, no updates, and then in the same day you may have two or three.

Kim: Does it tell you, "by the way, you should reload me."?

Betty: Yeah. There's like a little thing that pops up on the bottom corner that says relaunch new version.

Adam: Yeah, it's a very persistent ask. It's not like a, "hey, if you want," it's like, "no, you should do this now."

Betty: Yeah, yeah. So I see my frustration more as it's part of the learning process. I'm not mad. It's just something I don't know yet. And until I learn it enough to then think about how I can use it for myself or what I need to change about my prior way of doing things to take advantage of this new thing, that's the foundation. Because that's just me being human and trying to change.

James: Can I ask you a question, Betty, about people telling you no? So we, on another podcast in this series, have discussed the fact that really, sometimes it's good to say no. So I'm a parent, so I definitely appreciate that there is a value in saying no to people.

So here's the thing. In the sense of Progressive Delivery and trying to understand. I mean, there's an aspect of just because you could doesn't mean you should.

Betty: Mmhmm.

James: And in this atomistic world that we're potentially moving to, where it's like, "oh, I'm just building the app for myself," I mean, does the model ever say, "no, Betty, that's a bad idea. You don't need that." Or are you continuously being glazed and therefore you're not in a position to have a bit of constructive feedback about what would be valuable in the app?

Betty: Yeah. I'm going to guess that was Kent Beck. The most interesting thing that I heard him say, that was recently. There's been two things I remember from him speaking in recent memory. One of them was, this is not pair programming, because the system can't tell you no. And that's really not pair programming. And I agree.

And the other thing was, you're more creative when someone tells you no, because now you either need to solve the problem differently, or maybe that wasn't a problem.

James: Our kids absolutely are great. As soon as we say no, they find a way around that, right?

Adam: Haha!

Betty: Oh, my gosh. This is how I learned that you know, kids already knew about proxy servers and all kinds of other stuff. Yeah.

Heidi: And they'll teach each other.

Betty: Yeah. Haha.

James: Oh, yeah. Kids and InfoSec and LLMs are-- That's just one whole spectrum. They're the same. Or rather, it's not a spectrum, it's a singleton. It's the same thing.

Betty: Yeah. Haha.

James: You know, we think LLMs are good at identifying how to get past our network controls. But, yeah, kids 100%.

Do you ever find yourself being like, "ah you know what? I've gone really far in a direction that maybe I needed some guidance."

Betty: Yeah.

James: You know, Betty, do you ever have bad taste?

Betty: Well, if it's personal taste for myself, does it matter?

Heidi: That's a really good point.

Betty: Like, if I'm going to just watch bad reality TV and be a trash panda, does it matter if it's just me having a moment to myself? No. But you ask a really interesting question, James. I think at work, yes, but I think that's what guardrails should do. I think in your personal life, as long as it's not breaking the law, I don't know why it matters.

Heidi: I think it's really interesting when you're like, "if it's not a product, does it matter?"And I think that's a question that we're going to have to struggle with in a Progressive Delivery way. Because if we get something customized enough, it's going to become so individual that it's unproductizable.

It's sort of like not many people wanted a hot dog stand theme on their Windows, but enough did that we had it. But what happens when we get the weird edges of taste and people want to preserve that, but it's not a thing that could be in the product?

James: This podcast should 100% have been called the Weird Edges of Taste.

Adam: The Weird Edges of Taste.

James: How do we end up with the Third Loop? The weird edges of taste is clearly.

Adam: It was right there.

James: It was right there.

Heidi: Haha.

Adam: It was right there. I mean, it's like you know, are we going to end up with Geocities 2.0? Is that what you're asking?

Heidi: Yeah. Well, and I think that Geocities was this flowering of creativity and also like, really horrific web design. But so many people, so many women I know in technology came up through Geocities because it was customizable, because they could make fan pages that were, okay, objectively hideous. But also the idea that you could change a webpage was the spark that lit their technology.

Adam: Well, I mean, it made the notion of the Internet at the time so accessible to so many people who otherwise had nothing. Right?

Heidi: Yeah.

Betty: Replace "Internet"with "software apps."

Adam: Yeah.

Betty: You know what I mean? I think for me, the defining line is, is it for, like a business or is it for your personal use?

Adam: But the interesting intersection there was-- There was a great talk at Monkigras this year we were all there and Danilo Campos gave a great talk on how he went and said he took this, some of the commercial equipment that he had in his house, physical Iot, I think it was a heat pump. And he rewrote the controller software for it because the stuff that was provided by the manufacturer wasn't actually doing what he wanted it to.

Betty: But for his use.

Adam: Completely.

Betty: That's what I mean. Yeah.

Adam: But like you know, then it's a function of like, okay, well it he open sourced everything, it's up on GitHub repo. So if somebody else takes that and uses it, it's no longer just for his use. I mean, sure, person's actively taking it, choosing to use it. But it introduces an interesting idea of like, is the maker of a physical thing that is using software, are they required to prevent their end user consumers from doing things? Are they Incentivized to or they are Incentivized not to?

Heidi: John Deere has entered the chat.

Adam. Yeah. Haha.

Betty: Yeah that's an interesting example. Haha. You know, they could, that manufacturer could say like, you're not allowed to run other software on our stuff because then you're out of support, as an example. But it doesn't prevent people from doing it, it just means like, hey, you're on your own now because you broke the terms of our contract and what you paid for.

I think the key is money changing hands because I think money changing hands comes with an expectation. It's an expectation that the business will provide the individual as well as like and so both sides are agreeing to a set of terms for that.

The fact that Danilo open sourced it also means that there's no change of money. "I have given you some tools. Go do with it what you will. It is not my fault if you do something and you fork this and you change it and your entire heating pump system breaks down. I did not build this for you."

James: Oh no, the people that didn't pay for shit, those are the ones that expect the most-- The most, the most.

Adam: Haha!

James: If you paid for it, you value it and therefore you think that the person that built it might have some vague value. If they open sourced it, you didn't pay for it, that you're like, where's my feature? Why have you not built my feature?

Heidi: I want to back up and say, hey, does anybody else remember a time when somebody made hardware work differently and then open sourced the results? Like maybe a printer.

Kim: Mmm.

Heidi: Ha. That's the origin of open source software is like, "I hate how this printer software works and I'm going to fix it for myself and then make an entire transnational movement out of it."

James: Although now what worries me, Heidi, is that we'll-- I think I just realized something. We're all going to get sued. We're going to have like, brother and HP all up in our business being like, why did you rewrite the controllers? Like, we--

Heidi: They can take it up with Richard Stallman.

James: Well, I mean, good. He would enjoy that conversation. But for the rest of us, heat pump rewriters, I mean, it's actually-- So Progressive Delivery. Great. Okay. We're gonna roll things out, get the Third Loop.

The Third Loop, bringing the user into the thing. Yeah, they're gonna rebuild shit and mess things up and have an infinite entitlement. But, yeah John Deere, that's a good example. I mean, tractors are not hardware anymore. They are subscriptions for support of the infrastructure. And you've just made me think about, you know, where Progressive Delivery meets the right to repair legislation.

Kim: Mhmm.

James: Because Apple likes you to have things when they want you to have them.As we've said, sort of Google and/or in the book, talking about like, Gmail or Google Docs or sort of Microsoft having a toggle where you're subscribed to the better experience. But if Betty's gonna build a new controller for her controllers for--

Betty: Haha. My heating system here or something? Yeah.

James: Yeah, yeah, for Betty's heating system. That's the thing. If we're not rolling things out progressively and making the user happy, then the user is just going to go and rewrite the shit themselves. Though not everyone is going to be like Danilo Campus. You know, we're definitely not going to be. Like most users, you know, it's not like, does this application work? You know, I need to rewrite it in C.

I mean, that's a whole-- But on the other hand, I do think there was a question about if we don't meet the user where they are, their ability to go elsewhere now, the Third Loop can be cut off and taken somewhere else in a way that was just never even remotely possible before.

Betty: Yep.

Adam: But I think that there's also the-- You know, you look at it, I think open source companies have actually had some successes in this model where, you think about it in the context of one of the things that I think successful open source companies have managed to do is engaging with those members of the community who do want something that is not built yet.

And whether that's being willing and able to take their submissions and feature requests or things that they've built on a fork and being able to merge them back into the mainline. There's an opportunity I think as well to actually use this as a mechanism for more deeply engaging with your users.

I don't think it's necessarily straightforward or easy, but I think that if people are enthusiastic, you know enthusiastic technologists that want to engage with your software and say, hey, look, I built this thing because it was, I love your software base package, but I needed it to do this one other thing and it might be useful to other people, but here I'm giving it back to you to just add to your entire platform for anyone to use.

Heidi: So this is like the games mod pack. Like okay, somebody built a whack ass thing for a Kerbal Space Program that does like real physics. That became part of the community conversation and spawned its own modification of how the Kerbal Space Program people thought of themselves.

Adam: I think that it's something that there will be people who figure out how to do this. There will be those that don't, you know. Absolutely. But I think that this is one of those things where all of a sudden it opens up such a tremendous opportunity where it used to be, that for you to contribute to open source, you had to be a developer, and that meant that open source projects that were geared towards developers or building things like developer tools had a ready supply of people that were willing to offer up time and effort.

If you expand that to like contributors, that is anybody who has like a desire to use your software for whatever task they need suddenly has the ability to contribute, that's potentially a pretty huge opportunity.

Betty: So I what you're saying is it sounds super interesting. There could be companies that say here's our thing, like we write our software, we do this. Could they make some element of that, like somehow publicly accessible? But I'm going to challenge you not on GitHub, because the average person doesn't know what that is.

When I first started using GitHub, I did pull requests to myself because I didn't know how all the mechanics worked. And you know what, everyone just talks in shorthand, if you're asking engineers for help.

If there is some magical way of like, like a GitHub type of thing is on the back end already and you're exposing some of that through this chat interface for people to say like, "I would love it if here it did this."

Heidi: Mhm.

Betty: And that's a way of getting-- Because now the best part of what you can do with these new tools is like, it's like building the UI part, right? Because that's what actual like non technical end users care about. They're like, I want it to do this, it's all UI related things.

If they could like submit things, suggestions that way, I don't mean write a whole thing. But just like it looks like this, I would love it if here it did these three other things for me.

Adam: I mean, I think that we are a few hours from the release of this podcast away from having that be like a pre-configured skill in Claude Code. Right? I mean it's not a complex thing for Claude to come or some equivalent interface to come back and say, "oh, you know do you have a GitHub account that you like already own? If you don't know what GitHub is, just say no and I will go, and I'm going to go create a GitHub account, I'm going to go do all the things to the appropriate repos,"without having to tell you what it is.

Betty: Why do you need the user to ever go do that?

Adam: No, that's what I'm saying. I'm saying that Claude would go do it for you. Right? So like because of the way GitHub in particular is set up, you could actually create all of the account information that you need programmatically on the back end under the covers without ever having to involve the user.

Now it raises the question or concern that I think GitHub's already facing recently where, all of a sudden the scale of their system is at a point where it wasn't designed for a few hundred million extra users to be spontaneously creating PRs multiple times a minute. You know, it begs the question of like, how are they going to continue to scale that interface and that programmatic access in a way that keeps it at least with two nines.

Heidi: Two nines would be fancy.

James: I think now was your opportunity, Adam, because today you're wearing my "this is my sarcasm"font T-shirt and I'm not sure you've really hit the mark, I'm afraid, on the podcast today. So maybe you could like reframe the two nines in a sarcastic way so that the sarcasm font could take effect because you've been, you've been quite earnest.

Adam: Haha!

James: So Adam, what do you think about GitHub's uptime of late?

Adam: I mean, I think that the five eights uptime that they've managed to maintain has been quite impressive over the past month. Haha.

Heidi: Haha!

James: I mean, agents are terrible things. They generate load like a-- Oh no, I was about to go, I was just going to do the F word there. That was going to be a, we were off the range of a family show, my friends.

Adam: Haha!

James: But yeah, it is really hard when, to your point, the amount of just the sheer scale of work being done by agents and the backends needing to process that. And yeah, GitHub is definitely in the front line of that.

Adam: And I mean, I think, look, I want to be clear, like sarcasm aside, huge HugOps to the GitHub folks that are trying to like, deal with this. We're talking about a rather significant step function in the order of magnitude of traffic that they are having to manage and mitigate.

And you know you talk to people who are on the open source maintainer side and they're like, "yeah, and the worst part about it is that the vast majority is complete crap and trash." Right? It is hot slop that I do not even want, but don't have the tooling, don't have the mechanisms to be able to say, "oh, well, I only want it from agents I trust."

I mean what does that look like? What is it? You know, So I think that there's, there are so many things that like we've talked about before on this. It's like--

That technological jerk is making things so fast and all of a sudden we're in this situation where we don't have the right tools and I don't think we even necessarily know what the right tools are for some of the changes that these new aspects of software have created for us.

Heidi: It's like cars need windshields now because we're going so fast that the wind makes it hard to see.

Betty: Heidi, your analogy is great. And it's because it's exactly like we have this one thing that's like, wow, but then all the infrastructure around it isn't there for this era. So it's not just about like, hey GitHub, HugOps.

It's like everything is going to get strained. It's like we have these tiny little pipes and we just opened the dam. And we're like, why is everything flooding? Oh, because our pipes are like only two inches. You know what I mean? We weren't ready if we weren't set up for this. So it's a whole lot of everything.

Heidi: Can we ever be set up for change that big?

Betty: No, I don't think so. It's like saying we can predict the future, then maybe that innovation wasn't as big as step function. I'm going to say like Internet's like this idea of the Internet and all this online shopping. Right?

Heidi: No one was ready.

Betty: Literally the logistics wasn't ready. You know what I mean?

Heidi: Which is fascinating because like Sears had been delivering fricking houses by train, by mail order. Like a huge portion of the country was getting stuff from Wells Fargo and Sears. And then we deprecated that infrastructure strongly and then we wished we had it back.

Betty: But even with getting houses delivered, so you large capital things to then say like you're going to now get like your pet supplies delivered daily in small nuggets, that dramatically increased the volume.

Heidi: Yeah.

Betty: And then I think with, even if we had like-- So say that we had a national infrastructure for that, the house delivery, they probably weren't set up at the volume.

Heidi: No.

Betty: Because you can't deploy all those rigs that would have been delivering houses to then dog food one day, like cat food the other day. Like, you know what I mean? That's just too much.

Adam: Right.

Heidi: But we did have like morning and evening newspapers. We did have two or three times a day mail service in the cities. Like I think we did have the volume capacity for that at one point, but the car revolution changed that.

And then because of that structure, now we need to rebuild delivery back in it-- Anyway, I just think everything, time is a flat circle. And I think that we're going to look at this current surge of AI as sort of like the permanent September of the Internet, which is a thing that old Internet people talk about.

It used to be that there was this huge surge of newbies to the message boards in September when people got to college and got their first Internet account. And there was like mass misbehavior all through September because people hadn't learned the community norms. And when Internet sort of democratized and became not just the venue of college students, it was called the permanent September because there were always people who did not understand the community norms in a community.

And I think that in the same way this, this AI revolution is like the permanent September of app building because there's a lot of norms that we expect from apps that we consume and we're really offended when they don't exist, but the people who made them didn't know they needed to.

James: Oh my God. That's another whole book that you've got there, Heidi. So that's intense. But yeah, this is the first time that I, since we began this journey when I started thinking about Progressive Delivery in the context of Amazon Prime.

Do we deliver in chunks? Do we get a thing and say we don't want this in a way that we can send it back? Do we end up with, like me, a bunch of things that we meant to set back and then didn't actually send back?

Betty: And they're by the door. Haha!

James: Oh, yeah, no, I'm 100% gonna return this item.

Adam: Yeah.

James: And then six months later you're like, "what is that item doing there? Why did I not return this item?"

Heidi: But it lives by the door now. It's just my pet wrong printer cartridge. Haha.

James: Exactly. It's the place where we put the envelopes.

Heidi: But also do we have an Amazon delivery day? Like, I don't want things coming in a trickle. I feel bad about that, like environmentally, or I can only be home I only want things on Tuesday. Okay.

James: Yeah, I've got a whole thing thinking about Progressive Delivery meets physical delivery.

Adam: Well I mean, you look also at examples like Amazon in terms of their infrastructure processes and how they actually will do item movement based on your search history. So like, even if you don't buy something, there's a good chance that they're actually sending an item to a local delivery area, a local warehouse, so that it's ready for when you want it. And you can get same day shipping.

Or you know, this goes back to you know one of Heidi's favorite talks that she used to give on disaster preparedness with Waffle House. So for those of you that aren't familiar, Heidi, why don't you explain just briefly.

Heidi: So Waffle House has an emergency response center and their own weather forecasters. And it's all extremely sophisticated. And what they do is they station trucks outside the blast radius of the hurricane that's incoming so that as soon as the road's clear they can resupply stores that were in the zone of destruction. And they have on call teams that serve the same purpose. They are prepared to make sure that as soon as possible they can feed people warm food.

Betty: They're famous for this.

Heidi: They're famous for it.

Adam: Unequivocally better than FEMA at this point.

Heidi: Yes.

Betty: No, they're famous for this as the outcomes that they deliver on this.

Heidi: Right. And it's very valuable to them as both a community thing and I think it's part of their values but also they're investing in this proximity because they know where they're going to be needed. In the same way that like maybe Amazon ships things and in the same way that if you're a local hardware store in the summer you stock screens and mosquito repellent and in the winter, okay, in my area, you stock snow throwers and sidewalk salt.

Like that's what's relevant to the people at the same time. Nobody's going to buy sidewalk salt in July. It's not a thing. You don't pre-stock it. But the delivery thing that I think is interesting is when you read a Victorian novel, people are having near real time conversations by postal mail.

Like you'll send a note in the morning and someone will send you a note back at noon and you'll meet in the evening. And that like super rich communications in the physical space is fascinating to me because it's almost impossible to do now and costs the earth.

And so what is it that we are not looking at that cost the earth? You know, if we lost email, could we function? Like lots of teams are like, oh yeah, we have Slack, it'll be fine. Would it really? How's your paycheck come in?

Adam: Yeah.

Betty: I will say that for the last two years I have barely looked at email at work. I had to put a note to remind myself to look at email because everything was in Slack.

Adam: But you know, you think about that in the context of communications outside of work.

Betty: Yes, that was where it fell down. Like you couldn't talk to anybody that didn't work inside your own company because you would forget.

Kim: So I've been spending a lot of time in Mexico and I do not like WhatsApp. But if you want to do anything down here, you gotta be on WhatsApp. You want to order a pizza, you want to talk to Friends plan a party, get the news half the time, that's where people are.

Betty: Yep.

James: Co-signed, European. Europe is 100% WhatsApp.

Kim: Yeah, it's not a choice.

Heidi: I wonder if there's a better skin for it. Like, I think what I hate is the way that it organizes messages and maybe I just need to break my soul to it.

Adam: Either that or you just need to ask Betty to code you a new interface for it.

Heidi: Yeah.

Kim: Wait a second.

Betty: No, no.

Heidi: No, Kim's going to do it.

Betty: I'm not in the business of having customers.

Adam: Kim's going to do it. That's going to be Kim's vibe coding projects.

James: Create a new WhatsApp interface.

Betty: Yes.

James: Do it.

Heidi: Yeah I just want the interface. Like, I want the underlying message protocol to stay the same, but its message threading is pants and I don't love it.

James: Okay. You know, we don't want to use WhatsApp, but Kim's going to create a new front end. But yeah, Betty, I mean, let me ask you as a seasoned community builder, marketer the world is changing so dramatically in terms of user expectations, in terms of platform capabilities, in terms of who the developer is. How do you see things?

And look, if you were even able to predict the next two weeks, that will be amazing. But what are you excited by and how do you see some of this stuff playing out in terms of just the new ways that we can deliver applications and services to users and maybe what the industry is going to look like going forward?

Betty: For me, it's a bit of like there's this-- You guys have Progressive Delivery, and I think that is a, a way to think about broadcasting out as an organization. Broadcasting out. I think of this user empowerment movement, that's what I think this is, in that it is like progressive consumption.

I get to define what I want to use and when. It's a little bit of like, I wanted to say it's like the gig economy of apps, but that's not quite it, but it's a bit of like a company, corporation, business. You don't have to dictate to me exactly how I do this thing, what I use to do this thing.

I may still like the thing that you do because I need the chassis, I need some percentage of the way there, but my progressive consumption of that is going to be much more individual. It's like the parts of the Venn diagram of the ICP that do not overlap and I should be able to have that.

Adam: Right.

Betty: And maybe that part of the burden can be mine, doesn't have to be yours. Because that does not make sense as some sort of like scale economy for a business at all.

Adam: Right. If you've got a million users. It doesn't make sense for you building bespoke offerings for every single one of them.

Betty: Yeah, it's a little bit like I was looking up gig economy the other day. And it's part of that definition, but in a slightly different way. I think it could still be we have opportunity to come up with a name because you gotta name it for it to be true. I did see that on one of your other transcripts.

Heidi: Yeah. I think that it's a personalization world. It's bespoke to self.

Betty: Mhm.

Heidi: And there's a difference between tailored for you and bespoke, which is like a tailor is opinionated and will make changes that make the you inside the suit appear different on the outside.

A bespoke suit is just, it's your measurements the way that they are presented to basically the machine. And what we're saying is we can't all afford tailors, but maybe we could all afford bespoke, where something fits us properly instead of us having to adapt ourselves to fit it.

Adam: That's deep.

Betty: Mhm. I also see this as like a, "I get to define what it is."

Heidi: Yeah.

Betty: Versus like someone else tells me that this is what it should be. Yeah.

Kim: It's Geocities and MySpace all over again.

Betty: Differently. But yeah, all those led to something else. Right?

James: That actually makes me feel really good about it because so many people in this industry, people that didn't come from a normal computer science background, got their backgrounds in Geocities like that was-- Yeah. And again, MySpace, like the introduction to web design and maybe a less gendered cohort.

So man, I'm really wanting to drop the F bomb, but I'm definitely here for it. If that's where we're at now, then that gives me much happiness and optimism for the future.

Heidi: Yeah.

Kim: Betty, who else should we talk to?

Betty: So I was thinking about this and one, I have a name for one person and two, I have a second recommendation, but I think it's-- I don't know who, but I think if you all know someone in those organizations, that would be important.

First my good friend Laura Tacho. So she's at AWS now, but before she did a ton of developer productivity, like research and then also, you know her work at DX and then worked with a lot of these engineering leaders about like being more like effective on output. Right? Delivery. So it'd be super interesting to see how those organizations think about that third loop of it and is it changing from her research and all of her work with those engineering leaders.

I think that would be interesting. And then the other one is I think you all should talk to somebody in a consumer company because I think so much of like all of us where we have worked at is like tech companies for selling B2B tech. The mindset on that is super different than B2C, especially that third loop of who that is. Imagine if you're like, I don't know, you build like a social app. It's anybody.

Like I listened to like the opener of Lenny's Podcast. That's as far as I got because I don't always listen to work related things. But he had the founder of Snapchat on there and he just gave like a quick little bit and I was like, oh yeah, you all should talk to somebody from like a pure consumer, mass market consumer company.

Because I think that would be super interesting on how they think about third loop. Do they? Someone like Snap was like, they didn't do the things the way that every other social platform did. And so what does that mean for third loop considerations?

James: Love it.

Kim: Nice.

Adam: Well, thank you once again. Really appreciate your insights and looking forward to seeing what you build next.

Betty: Yay. It was so lovely to chat with you all.

Heidi: Thank you.

James: Thanks Betty.