Aakash GuptaComplete Vibe Coding Tutorial: Build a Full Stack App with AI | Andy Carroll (Windsurf)
EVERY SPOKEN WORD
65 min read · 13,247 words- 0:00 – 3:47
Defining “vibe coding”: humans + AI building together
- AGAakash Gupta
What is vibe coding?
- ACAndy Carroll
I think it's really about humans and AI tools vibing together to create new products, new services, to bring new things to life. Business leaders overwhelmed with the amount of noise that they're hearing about AI, but many struggle how huge a time saving this is. But to me, this just feels effortless, right? It's click a few buttons and, um, you do need to be really kind of careful though. I've actually been working in product for about 16 years and, um, didn't come from a technical background. So the right answer on the internet is to post the wrong answer.
- AGAakash Gupta
Yes. [chuckles] The fastest way to get a good design is to put a bad design in front of your designer.
- ACAndy Carroll
Exactly. When we start out, right, let's define what it is that we're actually trying to achieve, what is gonna make this unique, and let's focus almost all of our energy on making that part of the product really, really great.
- AGAakash Gupta
In today's episode, I sit down with Andy Carroll, a PM of over 15 years, and we walk you through how to vibe code. We'll specifically use Windsurf and show you how you can deploy websites, create mobile applications. If you have been interested in learning more about vibe coding or Windsurf, and if you are a PM interested in understanding whether you should be vibe coding or using Windsurf, we will cover all of that. So without further ado, let's get into today's episode. Really quickly, I think a crazy stat is that more than 50% of you listening are not subscribed. If you can subscribe on YouTube, follow on Apple or Spotify podcasts, my commitment to you is that we'll continue to make this content better and better. And now on to today's episode. Andy Carroll, thank you so much for being here.
- ACAndy Carroll
Thank you, Aakash. I'm extremely excited to be here.
- AGAakash Gupta
We're both longtime PMs. Neither of us are particularly technical in our last few roles, and so today I wanna break down something that a lot of people in our position might be looking at, vibe coding. What is vibe coding?
- ACAndy Carroll
Sure. Um, I think you've hit the nail on the head. Um, I've actually been working in product for about 16 years and, um, didn't come from a technical background. Um, I'd describe myself as, uh, technical enough to be dangerous. Um, but, um, you know, to be able to have an informed conversation with engineers and with designers and to understand how things are built, but certainly not to really get stuck in and actually write the code myself. And, um, I think one of the most transformative developments, uh, in the AI space over the last, you know, few, few months and years has been the introduction of these AI-powered, um, integrated development tools, uh, like Replit and Cursor, both new, Lovable, and the list goes on and on. And, um, vibe coding is the process, uh, I think, uh, uh, I think they may be drawing some inspiration from, uh, the jazz kind of movement, but, um, I think it's, it's really about humans and AI tools vibing together, uh, to create, um, new products, new services, to bring new things to life. And, um, so certainly, um, although the term does make me cringe a little bit, um, for me, it's, it really is about, um, making, making something and really just sort of seeing where it takes you. So a lot of the time it's, um, planning enough to get to the next step and then really being open to exploration, and this is where I think these AI tools are just, you know, unbelievable.
- AGAakash Gupta
Yeah. I believe the term came from Andrej Karpathy. He was early at OpenAI. He worked at Tesla AI. He himself is pretty technical, but he talks about this process of vibe coding, where he just lets the tools do most of the coding. And so that's what I wanna walk people through today. What is your tool of choice for vibe coding?
- 3:47 – 5:21
Tool landscape and why Andy chooses Windsurf
- ACAndy Carroll
Sure. So like I mentioned, there, there are many, and I've probably forgotten quite a few. Um, and as I said to you earlier, you kind of have to, uh, pick your weapon of choice. Um, the one that I use most heavily is called Windsurf. Um, I think the company that built it was called Codium, and I-- they may have rebranded as a result 'cause Windsurf has become the, the flagship thing. Um, but it's, um... I suppose w-when I think about these tools, there, there are a number, um, and they sort of hit different kind of moments, I think, in, in someone's, um, exploration sort of journey. Um, starting out, uh, I think one of the, um, one or two of the b- the better ones for, um, getting started quickly are probably v0 and Lovable. Um, they have really, uh, amazing-- I mean, Replit similarly, and Replit's got an amazing kind of mobile, um, experience as well. Um, those are very easy places to start and to get something, you know, to do to put together a quick landing page. Um, they've got an amazing community section where you can draw inspiration from other, other things that have been built. But as soon-- W-Where I found myself a little bit limited was actually wanting to get into the full kind of stack and see exactly what was going on throughout, um, sort of the code base, um, and I suppose Windsurf provides more of that experience. So it's more of a full-powered kind of coding editor with an AI copilot, um, that is also, um, pr- which also provides access to all of the leading, uh, LLMs.
- AGAakash Gupta
Awesome. So can we see this in action with a little bit of a live cooking session?
- 5:21 – 7:53
First look at Windsurf: workspace, files, and Cascade copilot
- ACAndy Carroll
Definitely. Um, so a little bit of background, uh, quickly, um, before we do that. Um, so I'd say I created a GitHub account, uh, about, you know, like I said, fifteen, sixteen years ago, and, um, there wasn't a lot to see in my GitHub account except for a couple of cloned repos and a few other things that people had shared with me. And, um, then just over the last sort of, uh, three, four months, and mainly the last two months, um, the, the number of things that I've started doing has really kind of exploded. Um, and this is all thanks to, um, to, to live coding. So let me share my screen with you. Cool. Now, looking at this can be a little bit overwhelming. I think, um, before I started doing this, I kind ofYou know, I'd, I'd used VS Code as well, and so had some familiarity there. But, um, certainly looking at this was very much the same to me to begin with as looking over a developer's shoulder and seeing, um, you know, loads of complex things, file structures, as you can see over on the left-hand side, and, um, and various other things. But actually, in terms of a, a simple structure, what you have over here on the left-hand side is just a set of folders. These are just my sort of active or some of the active projects that I've got running that I've added to my workspace. So these are the things that I'm currently using, um, actively and, and developing. And like I said, um, to my point, I've probably got a few too many things going on, but this kind of gives you a sense of, um, just, you know, how creative I think these tools can enable you to be and also how much you can achieve, um, in a, a fraction of the time and with a fraction of the resources that was previously possible. Um, and so if you go in here, so, so what, what I've got over here is, um, a project for, you know, a, a few of the clients that I'm working for, as well as some of my, um, my private projects and sort of personal website. And then what, um, what Windsurf gives you is... So the copilot that I mentioned is called Cascade, and you can see over on the right-hand side, top right corner, um, over here, you've got this, um, this little chat widget on the right-hand side. So, um, I'll just close the, the, the conversation that was open there. Now, again, it looks like there's quite a lot to look at, but once you get the hang of it, it, it takes a little bit of
- 7:53 – 8:54
Chat mode vs Write mode: controlling risk and avoiding “overzealous” edits
- ACAndy Carroll
getting used to, but it's actually pretty simple. You've got "write" over here, and as it says, when write mode's turned on, Cascade will be able to make changes to your code. This is crucial. Um, I think, um, I [chuckles] I didn't realize that this existed for about the first 40 hours that I used the tool. And, um, one of the things that I would say is that, uh, these tools do have a habit of being slightly overzealous in, um, their, um... Le- let's just say they're very proactive. [chuckles] So you could give it a, a prompt that said something like, um, "Hey," like, "I'm trying to change the font color over there on this page," and then suddenly the whole page would break. And then I'd say, "Well, you know, why did you do that?" And, and, uh, it would just reply, "Oh, sorry, that was a mistake. You know, I just went in and refactored the entire landing page." So do be [chuckles] quite careful. But chat is a safe, a safe place. So chat basically switches off access to all of the live coding editing capabilities.
- 8:54 – 14:18
Model selection strategy: using different LLMs for different tasks
- ACAndy Carroll
And like I mentioned, you then have access, and I know with, um, with Cursor, there's a similar thing. You've got access to all of the, the big ones. What I have noticed is that DeepSeek is not available in Cursor, and, um, I don't think Groq is either. So I think that, um, what you have access to here, um, gives you, uh, a lot more sort of, uh, capability, a lot more variety. Um, one of the things that I do find is that different things are good for different sort of aspects of, uh, of a project, and this is where I think-- or this is what I found sets Windsurf apart from my initial experience with some of the other simpler tools. Um, you know, what you have here is the ability... So for instance, you can see that you've got Claude 3.7 Sonnet, the thinking model, which is slightly more expensive, and then you've got DeepSeek-R1 as well. Now, both of those models are really useful, um, I find, when you wanna go broad with something. For instance, if you just re- you really wanna kind of... You wanna think something through without getting to a final answer in one shot. So often, if I'm sort of trying to establish context, um, you know, saying, "Okay, well, I'm thinking about building this thing. Help me think through the different pros and cons of various tech stacks that I might choose." For that kind of thing, going broad is really useful. And what I've also found really amazing is just sort of seeing, um, how these things work, right? So as you're doing that, then let's, you know, let, let's, uh, use that as an example. So if I pick 3.7 Sonnet and start having a chat. Um, "Hi there." Um, I... And I'm, I'm still probably overly polite. I think, uh, there's something about, um, you know, I do say please and thank you, but maybe that's 'cause of my fear of, uh-
- AGAakash Gupta
Today's episode is brought to you by Maven. The problem with most courses online, like Udemy, is there's no live component, and the instructors aren't experts in their fields, they're professors. At Maven, you get direct live access to experts and operators from the world's best tech companies. You can't get that access anywhere else, in any university, and you usually can't find them on YouTube either. I've featured so many of Maven's experts in the newsletter and podcast for that reason. To help you out, I've put together a collection of courses I recommend at maven.com/x/aakash. This includes courses like AI Prototyping for PMs, Product Sense for PMs, and getting an AI PM certification. Visit it now at M-A-V-E-N.com/x/A-A-K-A-S-H. Today's episode is brought to you by Miro. Let me ask you something. How many tools are you juggling just to get a single project across the finish line? One for brainstorming, another for planning, something else for tracking tickets. That's where Miro comes in. It becomes an all-in-one collaboration workspace. Whether you're consolidating user research from several interviews, developing and synthesizing product briefs or a wireframe, or project managing development, Miro brings everyone into the same space. It's fast, intuitive, and fully loaded with features like project templates, two-way Jira sync, and integration with software like Draw.io and PlantUML. Miro's AI features can be used to synthesize elements in a board to develop a ready-to-review product requirements document in seconds. If you're tired of tab overload and scattered workflows, try Miro. Head to miro.com and see why over ninety million users choose Miro to guide from idea to outcome
- ACAndy Carroll
The robots taking over at some p- at some point. But, um, [chuckles] I, uh, wanna start a new project, um, and, uh, f- and, um, uh, want to build a full stack, uh, web app with mobile- Uh, yeah, corresponding, uh, mobile applications. Uh, talk me through.
- AGAakash Gupta
So you wanna start a new project. You wanna build a full stack web app with corresponding mobile app, and you're gonna ask Talky to-
- ACAndy Carroll
Yeah, talk me, talk me through the process, uh, you know, uh, ask me, uh, questions. Now, typically, you know, um, you know, w- with the, the, the idea of vibe coding here is very much, you know, you can start with, uh, an idea, a very clear idea, and we- we'll, I think we'll get to one of those, um, in just a moment. But you can also start with something completely open-ended and just get, you know, and just say, "Well," you know, and, and what's probably gonna come back is, "Well, what do you wanna achieve? Who might this be for?" Um, but you'll see here, so because I've got Claude 3.7 Sonnet Thinking, you should see a thinking process there, right? The user wants to start a new project. Okay, cool. And then it goes back on itself, and it goes, "Right. Let's go." You know, what problem are you solving? Feature requirements. What are the three to five most critical feature requirements, technical considerations, et cetera? Now, one of the big things that I've also been doing, uh, recently is actually to, um, structure a series of, of prompts. I've got a very detailed, um, prompt library as well, which, um, I can, I can show you over here, um, which is, um, all around, uh, building out. B- basically, like, uh, the process of learning all of these things, some of them in really sort of painful and quite expensive ways, [chuckles] has made me get back to a lot of the basics of being a product manager. And actually-
- AGAakash Gupta
Sorry to interrupt. Uh-
- ACAndy Carroll
Yeah
- AGAakash Gupta
... can you hit "Don't allow" on that Riverside pop-up?
- ACAndy Carroll
Oh, sorry. Yeah, sure.
- AGAakash Gupta
Sure. All right. Let's go back. Yeah.
- 14:18 – 16:18
Planning before coding: prompt libraries and PM fundamentals
- ACAndy Carroll
No problem. So, um, like I said, um, one of the things that, um, one of the most expensive mistakes that I made actually when I started out using this was, uh, to dive in kind of headfirst with a bit of a, like, "Let's have a chat," and, "Cool, let's do that," and, "Let's build," I don't know, "Let's build a React app," uh, 'cause I'd used React quite, quite a bit, and obviously it's very popular and well-supported, et cetera. And I didn't actually run through some of the planning sort of process with any kind of real methodical rigor, and that really came back to bite me. And actually, you'll see, like, I, I may, may get a chance to show you, but I've, I've had sort of two or three false starts with that specific project, where I actually ended up having to just start again. Um, and so what I've done now is actually built out a, a very detailed set of prompts which basically helps you, you know... And this is the kind of thing that I'd love to sort of share with, uh, with you and your users and, a- and, uh, your audience. But, um, it's really just, um, something that I think can, can help people to structure their project in much the same way. And this is what, when I think about, like, the value that this c- can provide, I'm thinking largely about not just sort of individuals trying to use this, but sometimes really non-technical, um, sort of business founders who may not have any history of product management expertise. A lot of the things that you can do with these, and this is where they really come into their own, is to actually, um, prompt it with, you know, "Help me, uh, help to act as a CTO and advise me on the ideal tech stack." You know, that's one example. Or, "Help me to establish a, a sound kind of product vision and strategy." And, um, you know, and then I've got a few different frameworks, some common ones, jobs to be done, user story mapping. You don't need to do all of these either. So, you know, you, again, you, you know, you can probably get lost doing documentation forever, which is definitely not the goal. Um, and Amazon PR FAQs, uh, you know, working backwards type thing. But the, the real kind of thing to me is actually that starting with context, like really clear context, and over-investing in the planning process of actually setting this up pays for itself, you know, many, many
- 16:18 – 19:21
Project example: AI Sports Reporter origin story and first prototype
- ACAndy Carroll
times over. So I'll get back to, to Windsurf. Um, what I'll do, I'll just bin that. Now, the thing that I've created today was actually, and, and this goes back to my kind of... One of my initial sort of origin stories with using these products was actually, um, w- two, two of my sons, so I've got twin boys, um, play football on the weekends. And after one of the football games, one of the coaches put together a match report and, um, did so just, you know, manually. Wrote a, wrote up a lovely match report, highlighted each of the players' contributions. Someone scored a hat-trick, you know, various other things, and shared that with everyone. And, uh, all the parents loved it, and someone said, "Hey, you should start a podcast. You know, have you got a thought of a career in journalism?" And I thought, "Well," you know, I was just kicking the tires with Lovable and a few other things, and I thought, "Well, actually, that could be quite a lot of fun." Like, why don't I, you know, uh, why don't I think about a, a little AI, um, sort of match reporter, which could be a thing that, you know, um, people might use? And, um, I can show you the, the sort of first, um, sort of version of that, um, which was, uh, this guy. But basically, what, um, what I then ended up building was a very simple, uh, match report generator. So this was just a, a simple waiting li- waiting list, um, idea, and then, um, basically you could run through, record your event, submit your details, generate a match report, um, both in terms of a, a written match report and then an au- an audio one. And this was really my first kind of experience of, of building these things and using them to basically, you know, do whatever you want. Um, and, um, like I said, I ended up not really doing very much planning, um, and, uh, and that really came back to bite me. But, um, now what I do... So as you can see over on the left-hand side, as I mentioned, you've got a, a series of, of projects and, and folders. And I created this, um, new sort of fresh version of this AI sports reporter, um, sort of project for us to use today.And one of the things that, um, as, as I've just mentioned that I, that I, I always do now is to produce a series of, uh, different documents. Um, you know, really what you call these things I, I don't think really matters. But, um, the common one is obviously that you have some kind of README file, um, which, uh, is, is really just a... A- and as you, as you can see, as you open that, it, it opens up in the middle. Um, let me just bring up Cascade on the right. So you can see that this is the typical kind of view that you have. So you've got code in the middle, um, you've got your chat on the right-hand side, and you can choose, you know, which model you're chatting to, and then you've got your file system on the left-hand side. And that really is all there is to it. Now, when you're chatting to, um, the LLM on the right-hand side, um, to, to the model, you're basically, um, going to be asking it to then make edits, code edits to these files over on the left-hand side. And as it starts doing so, it'll open the, the relevant file. But, um, like I said, starting out, so I, I now will start
- 19:21 – 23:27
Creating “one cockpit” documentation: strategy, architecture, roadmap in-repo
- ACAndy Carroll
with a series of sort of simple prompts. And like I said, I, I literally just did this yesterday. But, um, putting together a, um, a strategy document, you know, what are we trying to achieve? Who's this for? So AI sport reporter transforms everyday sporting moments into professional, shareable stories. Um, core product principles, value proposition, target market, all of these things. Now, when I think back to doing a lot of these things as a product manager, you know, these things often, you know, frequently take many weeks to draft and review and decide and all of those things. And this is certainly not replacing the process of, um, you know, it's not saying to everyone, "Hey, you should just go off and vibe code on your computer on your own in isolation and come up with a product vision." But what it is doing is enabling people to go from, you know, zero to 90% or 80% very quickly. And I think that to me is what's so exciting about these tools is how they are collapsing the feedback cycle, not just in, um, the sense of actually building new products but also in all of the stuff around these products. So in addition to creating a product strategy and an, and an, uh, project overview, um, I also then work through like what the architecture might look like. And to the point earlier about wanting a full stack, uh, web app and mobile apps, that was one of the things I didn't really think through sort of very carefully first time up. And I started building in React and then realized I'm either gonna need to go with React Native if I wanna have an iOS app or an Android app, whatever. And actually, one of the technologies that, uh, is also kind of i- in that, um, single code base, multiple app sort of space is Flutter. I've never used it before. Um, but um, you know, this is one of those, those opportunities. So actually what I've decided to do now is actually to use Flutter to develop, uh, this next incarnation, which, you know, i- is, is something that I'm, I'm planning to get live in the next week or so. Um, and shall I stop there? I- is there anything... Do you want me to just keep going?
- AGAakash Gupta
How did you develop these? Did you use AI to develop these? How much time did these take?
- ACAndy Carroll
So, um, certainly sort of heavy use of AI and then tweaking things along the way, um, given what I know and what I've done before. Um, there are obviously massive benefits of kind of knowing, um, how to, you know, what, how these, um, models and frameworks, you know, can be useful, um, which many who start out using these tools don't. And I think there's very well-documented, um, examples of people diving in and then just building something with AI and putting it live and then having massive security problems and various other things later. So, um, a process here of, um, y- basically sort of drafting some things and then chatting to AI and then building a repository of these things which is sort of developing over time. So as I come across new things or I go, "Should I really need to, uh, formalize that?" Um, I basically just sort of add to that over time. But as you can see, you know, the, it's certainly not mandatory for everyone starting out to have, you know, seven or eight different files over here. You could very easily have one or two. But what is key is that you start with something that really kind of clarifies what you're trying to do. And then another really big development, um, sort of recent one actually, um, for me has been to map out in a lot of detail, um, a product roadmap which then becomes a living document. Um, and again, you know, these are things that like frequently take product teams, you know, many weeks to put together, and this is something that I put together, you know, in just a couple of hours. But what you can also see is that I've been updating it as I've been going, and all the green check marks are things that I've done in, you know, just a few hours, right? And, um, and now I can see, okay, well, you know, where am I going, right? So in terms of what I've actually achieved, um, I've now set up a full, um, Flutter, uh, sort of project, uh, and structure. Uh, I didn't actually realize, and this is another big learning curve, I didn't realize how, um, how heavy I
- 23:27 – 28:45
Deployment realities: GitHub discipline, CI/CD, and why “deploy early” matters
- ACAndy Carroll
suppose the Flutter kind of development framework actually is. It installs a huge number of things. Um, here you can see all of these Flutter build files. Um, and I actually ran into quite a few sort of challenges with deployment. So that's actually another key lesson to, um, to really emphasize is that with v0 and Lovable, um, and a couple of others, I think Replit, um, for instance, the process of putting something together and then just one hit deploy is seamless and really easy. Um, you know, it, it, it, it's, uh, I don't know i- if it can go wrong, but certainly it's never gone wrong for me. Whereas one of the biggest kind of, um, sort of head-in-hands learning moments with, with this, with this tool was certainly that, um, not getting something live, not deploying something, not sort of committing code to GitHub regularly and then integrating with a hosting and deployment, you know, um, platform likeUh, Netlify is the one that I use, but there's also Vercel, which is a very popular one. Um, I've not used it personally. But what I would say is that getting a simple, um, I suppose, dev tool ecosystem together that you're comfortable with is a crucial sort of early step. Because what you wanna do is to get, uh, the, um, the simplest version of the system live and deployed, and then to iterate in complexity constantly s- sort of, um, increasing, you know, what's in there, but making sure that there are those checkpoints. And the mistake that I made early on was that I was making loads of development changes and seeing how things were developing locally, and then as soon as it came to deploy, I had, you know, just hundreds of errors. I spent days trying to get the thing live [chuckles] and, um-
- SPSpeaker
Wow
- ACAndy Carroll
... all sorts of refactoring. So, you know, again, this is also why I'm so passionate about this is because, um, I'm now teaching other people how to use these tools, and I see people running into some of these problems early on, but I can spot them very early and I can help them avoid a lot of the pain that I experienced. Um, but yeah, so, so the other great thing about this, and this is where I, where I think a little bit about how these tools are so profoundly impactful for, um, sort of product teams, is also that, um, if you use something like this and you... You know, GitHub and, um, deployment pipelines and various other things were typically the domain of sort of the, the software engineers. Like, I knew where they were, but I didn't spend a lot of time actually looking at them. I was more in Notion or Jira or wherever else, um, you know, uh, looking at, you know, how we were documenting things and visualizing what was going on in the roadmap and working through, you know, Figma designs and, a-a-and, and those kinds of things. Um, whereas what I've now been doing is, is putting everything into these file systems. So I've got this one, one system kind of cockpit, I suppose, for driving all of the things that are open. So as you can see on the left, I've got this AI sports reporter. I've now got all of my documents relating to, to this, including my prompt library. So I go in here and I say, "Okay, cool," you know, this thing's enormous. Um, but, you know, here, here's a set of different prompts that I might wanna use. What's my architecture? Okay, great. Um, product strategy, product overview, roadmap. Now, when you cross over to, um, to GitHub, um, I'll show you over here. You can see on the left-hand side that I've got exactly the same, obviously, you know, the whole repo, the wh- the whole thing's just been committed to the GitHub repo. But when you look at the roadmap, what's amazing now is that I've got my roadmap perfectly, you know, laid out in a ser- in a series of phases. Now, I could turn this into a PowerPoint slide deck or whatever the case might be, but the fact is that anyone working on this project and anyone in the team now could be given access to this, this code base, this repo. Anyone can look at this and make sense of what's going on, right? I know what you're doing. I know what the strategy is. I can challenge it if I want, but, you know... And here I can see, so here's what we're planning to do. You can see what's been ticked off, and you can see what's in progress over here with the little progress sign. So these are the kinds of things that, again, like teams spend... I've been guilty of it. Like, I know I've been, spent, you know, hours and weeks [chuckles] producing this kind of thing. And, um, and then what I've also done is actually created a, a little prompt, of course I have, um, which is, uh, to create a comprehensive project status report based on what you've achieved in the roadmap so far. And because you can run that in the context, so, so the whole kind of, you know, the whole system has all of the context for this project. You can say, "Review the roadmap, review the recent chat that we've been having, review all the files, and then provide a detailed, um, report for our exec team." And what I've actually done, and, and it's something we, we, we don't need to demo, um, here, but one of the things that I, I do now is that you can then take that detailed report, you know, here's an example output format, and then you can plug it into something like Gamma and, um, which is the, you know, the slide deck creation tool.
- SPSpeaker
Yeah.
- 28:45 – 38:52
Live shipping #1: fixing contrast by editing code via screenshot + Write mode
- ACAndy Carroll
And in moments, you know, instead of spending weeks on PowerPoint decks and whatever else, literally in 30 seconds, max a minute, you've got a beautiful detailed project summary for anyone's consumption, you know, to be able to share with anyone. Um, and so let me quickly show you, 'cause I'm sort of talking a lot about the tool itself, but let me show you the results. So this, and, you know, we, we can, uh, certainly, we, we can change this, uh, live, um, as we'd like to, but this is the, the thing that we've come up, that, that, that I've created, you know, a simple landing page. So step one with this project was, okay, well, we wanna build this whole multi-application kind of sport, AI sport reporter tool, but that might take us, you know, at least a few days, probably a few weeks. If I think about kind of what many companies do with really big teams, you know, most would look at that and go, "Okay, that's gonna be a few months," right? I sort of, I reckon that this is the kind of thing that might take two weeks, but the first step for me is just, okay, well, can we just validate whether there's actually something there? So step one is just let's build a landing page. And the prompt that I used to create this was really just, I wanna build a coming soon landing page. I want it to look sort of modern and slick and whatever else. Give it some kind of interesting background with, uh, some kind of, you know, CSS, uh, kind of treatment. So it's gone a little bit wild with the colors.
- SPSpeaker
Yeah.
- ACAndy Carroll
Um, but, but as you can see, so this coming soon, the red doesn't look great on, um, on, uh, on the background over there. So if I just... And, and this is the beauty of this, right? So I'll just say, okay, cool. Let me take a screenshot of that. Now, um, let me go back to Cascade. Um, where are you? Okay. Um, sorry, I don't want that
- AGAakash Gupta
Today's episode is brought to you by Amplitude. Building great digital products is hard. You know that better than anyone. Getting teams aligned, measuring what matters, and scaling your product strategy isn't easy. But what if you had a clear framework to guide your next steps? That's exactly what Amplitude built. They studied the best product teams to understand what really drives impact and turned those insights into the Digital Experience Maturity Assessment. In two minutes, you'll be able to see where your team stands and what you can improve to build better products faster. Click the link in the caption to take the free assessment and get a clear path to product growth.
- ACAndy Carroll
Um, here we go. Uh, so, um, back to, uh, AI. AI, uh, so AI Sport Reporter. Uh, please. Okay. 20-
- AGAakash Gupta
So you asked it to go back to AI Sport Reporter.
- ACAndy Carroll
Yeah, exactly. So, I mean, that's a, that's another key point is that, um, working with this tool, like what is really important is that, um... So as you can see here, now it's refamiliarizing itself with everything in, in the project. I have had, uh, and this again is something that, um, you know, probably beyond the scope of, of this chat maybe, but, um, getting into, um, managing the memories that, um, that the tool has. So there's a couple of different things that I, I didn't really sort of understand or really kn- kn- know or pick up existed early on. I knew that I had to provide a lot of context, but what ended up happening was it was auto-generating these memories of certain projects, and then sometimes when I was in the middle of coding a landing page for one product, it would just spit out a whole pile of content for something else, and I'd be like, "What, what?" You know what I mean? Like, literally, um, one of my clients is, um, an, uh, a fitness, um, a fitness business for, um, you know, prenatal and postnatal, um, sort of mum transformations. [chuckles] And that's quite different to, um, another project which is about, um, AI-driven, uh, talent acquisition, um, which we're working on. And you can imagine that those two things, uh, can create quite interesting results when they're mixed. So what is really important is that when people come back to this, they then, uh, refamiliarize the tool with what they're working on. Um, so, um, uh, let's look at the landing page. Uh, this, uh, coming soon, uh, doesn't look good on a red background. Uh, let me just change models here, like I said. Um, so I, I've typically used Claude 3.7 most of the time. What, what I have found quite useful actually, and one, one of the big benefits as well is sometimes when you get stuck sort of debugging something or whatever, for some reason, if you throw a different model at a problem that another one's been struggling with, it can actually... I, I think it uses the previous context and the things that the other model's tried, and then it, you know... So sometimes I find if I'm stuck with 3.7, I'll then switch over to, um, to 4.0, to OpenAI's, um, 4.0, GPT-4.0, and it'll for some reason just, uh, move forward. So those were-- That's one of those little sort of tips that I found, um, actually really helps. So let's look at the landing page. So as you can see, some of the models accept, um, sort of screenshots, uh, some of them don't. So let's look at the landing page. This "Coming Soon" doesn't look good on a red background.
- AGAakash Gupta
So 3.7 does accept screenshots, luckily.
- ACAndy Carroll
Exactly, and so does 4.0. So, um, let's examine the landing page. Okay. Um, yeah, okay. So let me just do something quickly. Um, don't, uh, change the background. Uh, please, uh, change the text color to a dark charcoal. Okay. Um, and as you can see, that's gonna be happening now. Like I said earlier, it's gonna... Am I-- Oh, sorry, I'm in chat mode. So let me just, um, let me just switch over. So key point there, it was trying to make the change, but it, it actually couldn't. Um, so make the change, uh, directly.
- AGAakash Gupta
So once we switch it over to write mode, it's able to directly change those code files.
- ACAndy Carroll
Exactly. And so what you'll see, the file opened over here, you can see the change. And this I think also gets to like the, the key excitement for me has been, as I mentioned, this kind of Cambrian explosion of learning. But, um, you know, I've actually just been learning, you know, by, by doing rather than something theoretical. So sort of seeing exactly what gets changed and how it works. Um, you know, I started out by, I suppose, leaning heavily and just letting the AI do all of the coding, and latterly, I've started making small tweaks myself kind of manually, um, and then getting the code to, you know, getting the AI to check that I didn't mess anything up. Um, and some of the time, you know, it's like, oh, I changed the title over there, and it'll pick up immediately, and it'll say, "You changed the title over there, but there's a JavaScript kind of, um, there's a JavaScript script that's, that actually picks up that title, the old title, and uses it. So the build process is gonna now fail." And so there's all of these kinds of things where it, it spots these dependencies, you know, not always, but, um, you know, it, it teaches you as you go what to look out for. And so often I'll go in now and I'll say, "Look, I'd like to change that title. Are there any dependencies or is there anything else I need to be aware of?"Um, here we go. So what's happened now, as you can see now, I haven't walked, walked you through the full kind of setup process. I've actually... Like, what I think is a, a better use of, um, people's time if they, they are interested, I've actually got a separate guide on some of that, and then also the Windsurf kind of getting started guide is a good place to start. Um, but, um, basically, like I mentioned, the, the major thing that you wanna do early on is make sure you've got a, a GitHub account, you know, free to set up and, and very easy and to create. I recommend Netlify. I've been really impressed with everything that they've done. But, um, and, uh, you know, we can share all, all of those details later. But basically, once you've got the W- Windsurf integrated with GitHub, um, you're good to go with a local kind of code repository and being able to push those changes to a remote repository. And then what you do is you set up Netlify to automatically pull. So you set up a sort of CI/CD kind of pipeline within, um, Netlify. And let me just show you that 'cause it should've started building now. Um, here you go. So you can see, you know, the range of different projects that I've got open. And like I said, um, you know, um, a couple of months ago, I didn't have anything that I'd built beyond, you know, a, a simple kind of personal website. And now, um, these are the things that are going on. You can see that within Netlify, um, you can see now that that change, which was pushed automatically to GitHub, has now triggered a build on the production pipeline, the production deploy here. And you can see that it's also added a nice little commit comment, "Coming soon text color to dark charcoal for better contrast." Um, and that's another really key habit, um, that I think is easy to miss, is sometimes you get so excited about making changes that then something breaks and you go, "When, when was the last time that I actually committed to GitHub?" Um, and that's something that, you know, really kind of I, I've, I've become fairly, um, sort of... I, I probably have over-indexed now on committing too frequently. Um, and I'm sure some in your-- some of your developers will, um, tell me that my, you know, my technique is not, uh, sort of quite perfect. But I find it very useful. So you can see here that, um, it's now initialized the... And there you can see
- 38:52 – 40:58
Local vs production testing + terminal inside Windsurf
- ACAndy Carroll
that "Coming soon" has been changed. Now, um, it's obviously a little bit of a wait to have to, um, see the full kind of deployment to production. Um, can you... So let-- I'm just gonna... What I'm gonna do is just ask Windsurf to bring this up on a, on a local, um, port. Can you get this up on, uh, local, um, 8000? I think that's the one I was using.
- AGAakash Gupta
And the difference here is what we saw was live on the web deployed via Netlify. Um-
- ACAndy Carroll
Exactly.
- AGAakash Gupta
And now we're gonna look at it locally.
- ACAndy Carroll
Exactly. So you can see over here, you know, and Netlify has just given it, um, a little Netlify app, um, but I've called it our sport report. Um, but what you can do as well with Netlify, which is amazing, um, is that you can, um, choose your own custom domain even on the free tier on Netlify. I don't know if there's limits as to the number of sites that you can use. Um, but certainly, like I, I don't think that that's offensive. You know, I wouldn't sort of, uh, not click on it if I wanted the product, but it is always nice to be able to have your own custom domain on something that you wanna look kind of really pro. Um, and, um, so let me just see if, um, that's come up. Okay. So another key point actually, which, um, i- is around the, you know, the idea of this being a, a fully integrated development environment, is the fact that terminal, um, you know, your, your, um, you can run terminal as a separate window. So, um, you know, you c- you can open terminal and, um, and it'll... Basically it'll, it'll open, um, your, your terminal view. But, um, what you can also do is, um, um, hit it directly from, uh, from within the chat. So here you can see it's come up with the command, and this just saves you having to use a separate product, right? So again, back to like just bringing everything into one place. Instead of jumping around between four or five different things, one for notes and one for strategy documents and the Miro board and whatever, it's just all here. And, um, from a coding perspective, it's very
- 40:58 – 46:31
Live shipping #2: generating and implementing a logo + favicon via prompts
- ACAndy Carroll
similar. Um, so running the background terminal command, so it's bringing that up. Um, okay. Let's see this. So, um, localhost 8000. Okay. So now we've got that. Now, when we make changes in here locally, obviously those will appear, um, immediately. So why don't we, uh, sort of take a look at this now. You know, I suppose, like if, if you were to look at this for the first time, um, actually o- one thing that sort of sprang out to me earlier was, um, one of the things that many product teams I think spend a huge amount of time doing, uh, w- certainly early businesses, is on coming up with kind of design systems and, um, brand logos-
- AGAakash Gupta
Yeah
- ACAndy Carroll
... and various other things, um, especially for new, for new businesses, for new products. And in my, in my mind, that's actually a really kind of low-value activity until you've validated that there's, there's actually something there that people want. Um, so again, what I've done is just created a simple, um, so sort of thinking about people who wanna start this out or wanna experiment with some kind of, you know, set up your own, um, sort of site. I've actually created a, a, just a little AI prompt, um, again, which, um, is somewhere in here, which enables you to sort of create a, a, a brand, um, logo. So a little bit further down. Um, but, um, so our prompt library brand logo. So help founders cr- quickly create a simple professional SVG-based logo and favicon. So let's just put that in here. Focus on creating something simple, distinctive. The logo should be professional enough to establish credibility, but can evolve as the company grows. Um, let's just put that in here. Okay. Um-I'd be happy to help you create a simple professional SVG. What's your startup or product name? Okay, it's already got all of that. Uh, you have all these answers. And just, you know, I would obviously, if I was doing this for a new business, I would spend more than five minutes or, or more than 30 seconds doing this. Um, you have all these answers. Yeah, so, okay, it gave me a little prompt. I'm creating a logo for AI Sports Reporter. Great. Um, so it's parents, color preferences. Cool. Using a gradient, style preferences, symbols or imagery, competitors. And you can see here that it's got some suggested actions, right? And again, back to, you know, if you're doing this on your own and you don't have access to a designer, or your designer's busy and actually, you know, you don't wanna wait two days to get something done, um, this is the kind of thing that en- can enable you to just get something good enough, uh, to start the, the process. Um, I mean, there's... A, a colleague of mine used to also say that, uh, the fastest way to get, uh, to the right answer on the internet is to post the wrong answer.
- AGAakash Gupta
Yes.
- ACAndy Carroll
[chuckles]
- AGAakash Gupta
The fastest way to get a good design is to put a bad design in front of your designer.
- ACAndy Carroll
Exactly. Um, so, um, uh, you know, I don't know. I like concept one. I didn't even read them, but, um, I'm sure it's good enough. Now, I'm still in write mode. I should be able... So it's gonna give me the SVG code for my logo. There's no, you know, I'm not running through some separate tool to do this. Um, you know, who knows what the result is gonna be. But it's given me that, and it's also given me a favicon. Um, it, you know, to be honest, almost gives you a little bit too much information. [chuckles]
- AGAakash Gupta
Yeah.
- ACAndy Carroll
But, um, but this is what's, what, what's been a lot of fun. So, um, okay, great. So, um, yes, please, uh, implement the new brand logo and favicon. Uh, the logo should be top left of the website and should, um... Yeah, okay, that's fine for now. I actually don't... It doesn't need to resolve to a homepage for now. Um, but I suppose, you know, the, the other thing that you end up having to do is to work through kind of navigation structures and that kind of thing. Now, um, okay, it's creating a favicon file, and that's being put into the, as you can see, static landing, and then it's created an assets folder. So it's doing all of this automatically. Right? Like, I'm not messing around with file structures and that kind of thing. You often do find that you have to start moving things around a little bit later, 'cause sometimes it gets, gets things wrong. Um, but I mean, if you just kind of think about... I don't know if you've ever had a similar experience where deliberation over what we're gonna call it or what the brand might be takes up almost as much time as talking about what the product should do. And I think-
- AGAakash Gupta
[chuckles]
- ACAndy Carroll
... that's, that's what you're trying to get out of really. Um, so here you go. The new logo perfectly captures the essence of AI Sports Reporter. The other thing is that these, uh, you know, [chuckles] they are amazingly optimistic at all times. Right. Let's see. So I'll just clear cache over there. Has it been applied? Ah, okay. Not sure that that's what I would've gone for. Um, but, um-
- AGAakash Gupta
But it's a logo.
- ACAndy Carroll
Exactly. So I've got a logo, and you can see that it's got a little favicon up in the top left, you know-
- AGAakash Gupta
Wow
- 46:31 – 54:01
Live shipping #3: adding a ‘How it works’ interactive infographic section
- ACAndy Carroll
... of the, of the, of the tile. Now, you know, what are some other things that I suppose people are gonna be thinking about? Like, if you were coming, coming to this fresh and going, "Cool, what is transform sports moments into stories?" Maybe we can, you know, um, co-create a little, uh, a little feature together.
- AGAakash Gupta
I think it would be cool maybe to have like a, like a infographic explaining what's going on, right? So I think maybe we show like input, recording of match, output, match report, something like that, that we put-
- ACAndy Carroll
Right. For the, um, home page, um, we need some kind of, um, how it works infographic that can walk users through, um, the high lev- walking users through the high-level process and what it entails, uh, step by step, uh, what to expect. [keyboard clicking]
- AGAakash Gupta
Okay. How is it dealing with our infographic?
- ACAndy Carroll
Yeah. So I mean, like, you know, again, it's sort of gone through... And, and this gets back to kind of if I was to do this with a team, right? Often this is just a, a great starting point for what are we trying to achieve here. Now, you know, capture the moment, input key details, AI magic happens, choose your format, um, you know, text, audio, PDF, and then share and celebrate, right? So nice, re- really nice little summary, right? Of, of-
- AGAakash Gupta
Yeah
- ACAndy Carroll
... of what's actually required in 10 seconds. Um, design considerations, right? Um, layout optionsUh, placement on the landing page. Um, free versus premium indication. Um, okay, now just for the purposes of time, uh, today, obviously we could spend a bit more time deliberating. Okay. Um, I'll take, uh, your steer on this.
- AGAakash Gupta
[chuckles]
- ACAndy Carroll
Sounds, sounds good. Um, let's keep it clean.
- AGAakash Gupta
And I see that little Accept All button. Is that another way to do it?
- ACAndy Carroll
Yes. So actually, so th- that's just telling me that three files have actually, uh, been changed. Um, so those were, um, it basically created the, um, the brand logo and the favicon and-
- AGAakash Gupta
Got it
- ACAndy Carroll
... there, there were a few, few other little things. So the index, um, file also changed, um, which you can see over here. So it's been modified, um, that little M over on the left-hand side, and that gives you a nice little visual cue of what's, you know, when something's changed, that a, a, a code change that may not yet have been committed to GitHub. And so when you start seeing that you've got, you know, 10 of those open, you know that it's, uh, it's probably time to, to push. Um-
- AGAakash Gupta
Yeah
- ACAndy Carroll
... let's keep it clean and simple for now. And, um, we don't have any graphics, um, at the moment. Um, so use, um, um... These are probably not my best, uh, quality prompts either, but, um, we'll see where we get to. I'll blame you if it doesn't look good.
- AGAakash Gupta
Quality of the prompt is everything, right?
- ACAndy Carroll
Uh, always. Um, I think this is also, like, why, um, the, the usefulness of having, like, a detailed prompt library is really powerful. Um, it's, you know, something that you can basically refer back to at all times, but just knowing that, um, you know, you don't need to think all of... There are established ways of thinking through many of these things, and, um, it's just so useful to be able to kind of do something like that, see what the output is, keep some of it, you know, delete some if you don't want it. But, um, it, it really is, uh, very powerful, um, as a tool. So it's generating that at the moment. Um, you know, I think what might also be useful is we'll add a little FAQ section or something like that down at the bottom, and then maybe we'll add a... We'll keep it to a single page. I don't wanna sort of introduce too much complexity now. Um, but we can add sort of anchor points to a hard work section and an FAQ section and something like that and, and maybe, uh, something else lower down the page. Um, now obviously sort of time doesn't permit us to go in and build, uh, the, the full kind of, uh, AI experience. Um, but you know, some of the key components I've mentioned GitHub, I've mentioned Netlify. Another sort of, uh, major thing is obviously to have some kind of, uh, backend, uh, database, um, system in place as well. Um, the tool that I've been using is Superbase. Um, again, there are a few different options. I've heard of a, a serverless option called Neon, which I've not tried myself. Um, but I think again, it's about sort of picking your weapon and getting comfortable with something. Um, you know, there is also the option to integrate with something like, um, Airtable as a very lightweight kind of, I suppose, less technical option. Uh, especially for something like a, a s- a, something like a, a sort of a content-driven, you know, marketing type site or whatever, where there may just be the need to update certain sort of content elements, but actually you, you would rather have something that's slightly more user-friendly from a CMS perspective. Um, that can be really useful. But if you actually want a fully fledged, fully featured, um, sort of database, um, structure with, you know, um, everything required to store user details securely and, you know, everything else, um, having a Superbase account is, um, i- is a really important step. All righty, hard works. Let's see what we've got. Looks great.
- AGAakash Gupta
So we're hitting accept on a few things.
- ACAndy Carroll
Accept, accept. Now these are just, so it's committing sort of, uh, these are terminal commands. Typically, you know, what someone would be doing without this is typing all of these things manually into terminal, um, I believe. [chuckles] Again, not having done this outside of, uh, this tool, uh, I'm not sure how huge a time saving this is, but to me this just feels effortless, right? It's click a few buttons and, um, you do need to be really kind of careful though. Um, and so this is another thing that I've learned is you do actually need to familiarize yourself with some of the key, um, sort of commands because some of the time it just recommends doing something that you immediately know you shouldn't, it shouldn't. Um, and some of those things can be quite difficult to undo if you get them wrong. Um, let's see what we got. Um, so how it works. Capture the moment. Take quick notes or photos during the game. Input key details. AI magic happens. Choose your format. Share and celebrate. Amazing.
- AGAakash Gupta
Love it.
- ACAndy Carroll
Yeah. So-
- 54:01 – 1:00:09
What these tools unlock: experimentation, team alignment, and monetization paths
- AGAakash Gupta
It's interactive too, so it's not just like random little things. When we hovered over it, it did it. So that covers our walkthrough of Vibe Coding. People have got to see us ship features live. We shipped a logo. We shipped interactive walkthrough of what the app actually does. If you were-
- ACAndy Carroll
Yeah
- AGAakash Gupta
... to summarize, what is the power of these tools and how should people be thinking about using them?
- ACAndy Carroll
Absolutely. So I think this is where I'm, I'm kind of most excited about all of this, is that, um, I, I'm passionate about helping, uh, people and teams to do more with less, uh, to, um, create, um, more. And I think a, a lot of people are thinking of AI as, um, something that's taking away, um-You know, human creativity. I'm actually passionate about the ability of, uh, these tools to make people more creative, um, to enable them to explore more ideas more quickly. Um, uh, you know, I've, I've worked in various product environments where, um, it may not have been a one-to-one mapping of ideas to production code, but sometimes it was pretty close. And a lot of that came down to the cost of experimentation, the cost of getting feedback. You know, there's a, there's a big overhead there, and there's often a big pressure to deliver. And what that often does is that you-- it, it sort of pressurizes teams to commit too early to an idea that hasn't yet really been explored. And some of the things I've shown you today, I think, uh, I hope have, have shown you how very quickly you can, you know... And I think the, the really big thing, um, to your question about, like, how are these things most useful, I think, is to align people around a common goal, um, really early. I think there are gonna be some barriers to, um, access, but getting people to use the same set of tools, I think, is really key. So, um, what-- where the benefit, I think, is gonna lie and where I'm seeing sort of, uh, or, or hearing of people facing challenges is that people are kicking the tires with certain tools, but you might have five different people in the same team all using a different tool to try and bring to life a new idea. Um, where I think the huge kind of benefits are gonna lie will be when teams can establish a shared mental model and playbook and sort of toolkit for how to use these things effectively, and that's where I think true transformation, um, can take place.
- AGAakash Gupta
Can somebody monetize an app with vibe coding?
- ACAndy Carroll
I think there's good examples of, um, of that happening, um, at the moment. Uh, there's, um, is it Peter Levels is a well-documented, uh, sort of celebrity, I think, in the, the vibe coding space, and there's, um, there are a few others as well. Um, you know, y- you spend a bit of time on X and y- you-- I think there's, uh, whole thriving communities. I think, uh, there's another guy named John Rush who I follow quite closely because he's doing some really great stuff, um, in this space. You know, building an ultra-lean, I don't know if it's a one-person b- business, but it's, you know, it's him and maybe one or two others, and he has, you know, 10, 15 different projects running at any point in time. And I think, um, Peter Levels similarly talks about how quickly he will trial something and then just with zero emotion just kill it if it doesn't work, right? And he just said that's one of the big things for him is actually, "I'm gonna go out with a, with a hypothesis. I might pivot here in the, you know, e- early on, but if it proves if people just don't really want it or I haven't quite nailed the solution, that's okay, right? And I can cut and I can reload." So I think yes is the answer. I think that's just one example, building products and getting them live, you know, micro-SaaS-type propositions where you can get something live in a weekend that solves a specific customer problem, just one problem, but does so really well. That's the kind of thing that people will definitely pay for. And then I think the other side of this is, um, doing, you know, largely what, what I'm focusing on at the moment, um, I think helping others to, to build these, to build these, um, these products and services quickly, and then also establish the processes around using them. So I think there's a few different sort of monetization, uh, options in this space. Um, and aside from that, it's also just, you know, addictively fun.
- AGAakash Gupta
Yeah. John Rush is coming on the podcast. He has agreed to come on the podcast.
- ACAndy Carroll
Amazing.
- AGAakash Gupta
I think his episode will come out, like, two months after this, so-
- ACAndy Carroll
Oh, nice
- AGAakash Gupta
... this is a little teaser for you all about the John Rush episode, and hopefully Levels is listening and we'll get him on as a result as well.
- ACAndy Carroll
Yeah, exactly.
- AGAakash Gupta
How should PMs think about using these tools?
- ACAndy Carroll
Good question. Uh, it's something I'm grappling with, um, at the moment. Uh, I think, you know, you, you really want, um, to empower people to, uh, explore things more freely, but also to focus on the stuff that actually drives differentiation and value. And I think when w- with a lot of these tools, um, what I'm finding most, most powerful and, and certainly something that I've, I think I've been guilty of in the past, is thinking that everything in the product that I was building was actually unique. You know, everything deserved the same amount of attention. And the fact is, you know, that's just not the case, right? And I think the way that I like to think about this is that you can almost remix and borrow and ch- a- a- and pick from, you know, choose, uh, draw inspiration from other things very quickly. So it might be, "We need a landing page. Let's not spend two weeks on it," right? There are 30 examples. I can take a screenshot of a great landing page, and I can get it live in an afternoon. If we wanna change it, that's fine. But I think, um, the ability to then say, "Okay, well, when we start out, right, let's define what it is that we're actually trying to achieve, what is gonna make this unique, and let's focus almost all of our energy on making that part of the product really, really great," right? And these tools, I think, enable you to, um, I suppose eliminate 90% of the, the busy work and the, um, sort of the drudgery so that product managers can actually focus everyone on the stuff that really is the, the secret sauce, the stuff that their customers really want.
- 1:00:09 – 1:12:01
PM career evolution, Andy’s journey, and building Aisle Partners
- AGAakash Gupta
So you've been a PM since 2008, actually coincidentally when I started as a product manager as well.
- ACAndy Carroll
Right.
- AGAakash Gupta
What shifts have you seen in product management, and do you think vibe coding will be one of those shifts?
- ACAndy Carroll
Um, yes. So 2008, uh, was an interesting year for me 'cause, uh, I got into product very much accidentally, um, by, um, losing my job in finance, uh, as a result of, uh, the trading business that I was working for, um, closing down. I'd been in London for 10 days. Um, so I'm South African, had moved here, um, and had envisaged myself working in finance, uh, you know, as a, as a trader. Um, and 10 days in, I was handed a cardboard box and told that, uh, the company wasn't, was no longer-
- AGAakash Gupta
Oof.
- ACAndy Carroll
Was no longer. Um, but they said, you know, "Do you wanna fly back to South Africa and, um, or, or do you wanna stick it out on your own and just sort of find your feet?" And, uh, so I chose the latter, and a friend of mine was actually working, um, at a music startup of all things, and it was kind of the early heady days, uh, post Napster, um, but pre Spotify and Apple Music and, and various other things. And certainly everyone knew that the world was shifting to a digital, you know, the, the music world was gonna be shifting to a dig- a digital model, but no one really knew how. Um, so he said, "Look, we, we are completely oversubscribed. Um, we've got, you know, everyone wants what we're doing. We're a small startup in, um, in West London and, you know, all of the world's, um, hardware manufacturers and mobile phone networks and everyone else wants digital music. So we've got way too much, you know, if you can start soon and, uh, basically, you know, without really having a job title, that'll be great." And, um, so I, like I say, like I, I joined a tech startup. I was still wearing a suit at the time, and that only lasted about a day because, um, everyone was wearing sort of hoodies and T-shirts and wondering who the finance guy was. And I sat down with, um, you know, various others i- in, in that space, and just within a couple of, um, a couple of weeks, I realized that this, this kind of job of working with... I, I was given, you know, an assignment to work with, uh, the design team on something, and then to work with the QA team, and then w- to work with the developers, and coordinating all of those things, um, to then ultimately bring a product to market was the thing that I learnt, um, there. And over the last 15 years, so I think early on, um, we didn't really have a, a job, a job description of product manager. Um, so my, my first, it was about six months in, I was actually told, "Well, actually the thing you're doing is kind of, you know, this thing called product management, and we have a new product that's gonna be going live and we think you'd be really good at kind of taking the, the reins on that." Um, and I think, you know, very big shifts sort of took place both while I was there, um, but, um, you know, the, the agile movement was actually sort of fairly, fairly new. You know, certainly something that, you know, w- was, um, wasn't something that had been formalized or, or, or certainly wasn't practiced everywhere, um, as it, as it more commonly is now. Um, but in terms of, um, real development, like I don't think that I've seen a shift, um, anything like what the potential of, uh, this new sort of vibe coding, um, kind of movement I think can provide. I think in terms of shifting people out of, um, a delivery-focused mindset, um, where, you know, everything is about measuring velocity and tickets delivered, et cetera, rather than sort of actual demonstrable, uh, value. Um, I think, yeah, I mean to, to answer the question, I don't know that there's been a shift quite like the one that we're seeing right now.
- AGAakash Gupta
Agreed. I think it's allowing PMs to get back to bare metal. You know, we spent so much time in the documentation and stuff like that, but it's nice to actually get back into the pixels, into the product.
- ACAndy Carroll
Yeah. No, exactly.
- AGAakash Gupta
So what are you up to today? You're running Aisle Partners. What is that?
- ACAndy Carroll
Um, so yeah, so sort of 15, 16 years in, uh, or after, um, building and, and leading, uh, product teams, uh, in various, um, sort of startups and corporates and a bit of, uh, time in consulting as well. Um, I sort of went out on my own last year. Um, so it's still sort of very, very early days, but really kind of focusing on everything we've been talking about today, but very much, um, so what, what we do is, what we offer is a combination of, um, highly targeted sort of AI, uh, training, uh, for, for typically mid-sized businesses, but a couple of our clients are, are quite a bit bigger. Um, and w- then we also offer a combination of, of training and then also implementation. So sort of agentic AI and automation, uh, tooling very targeted to building, um, I suppose you might call them sort of hybrid teams, but digital human kind of hybrid teams to help businesses, you know, like I said, do more, um, with less. Um, and I think, you know, a big part of this has been kind of, uh, in response to what we, what we hear really frequently is that, um, many businesses or business leaders are overwhelmed with the amount of, um, noise that they're hearing about AI, but many struggle to actually know how to apply it to what they're doing. And a lot of these tools and capabilities can also be introduced without requiring, you know, monstrous upfront sort of planning processes and whatever else. And there's a almost a bit of a collision of, um, how these things can be introduced in a matter of days or weeks to win back enormous amounts of time for an entire team. Um, and really like what, what we're hearing is a lot of businesses spend more time planning and talking about AI than they actually do implementing it. So that's, that's kind of what we've been set up to address is really, you know, help people kind of cut 90% of the discussion and the workshops and the massive sort of detailed upfront, upfront planning. And what we do is go in and say, "Right, you know, your specific team in your specific industry, you know, your marketing team, for instance, is doing certain things. Talk us through what they're doing. What tools are they using? What are they trying to achieve?" And then we help them very quickly map out how those processes can be improved with a goal of typically sort of winning back sort of two to four hours per person in the team within two to four weeks of, uh, engaging. And if you think about it, you only need to do that two or three times, and then everyone in that team has a four-day workweek, you know? And, and, and that's the kind of model that, um, you know, so the ROI on what we can deliver i- is massive. But really it's that combination of, you know, high quality training, targeted training, and then also implementation. And then the other thing that I'm doing sort of personally is more in product and digital kind of strategy consulting, all with a view to, um-introducing these tools, uh, to teams and helping them to get up and running fast.
- AGAakash Gupta
And what I've noticed is that the retention, the viewers who are retained to this point is something like 20, 25%. An abnormal amount of them are like us. They're thinking about, "Oh, should I do consulting? Should I stay a full-time PM?" How has that shift been for you? Have you been able to earn as much as you were earning before? Has it been as rewarding?
- ACAndy Carroll
Um, it's, I mean, really, really challenging, to be honest, um, just because it has been so, so different. I think, um, very early on, um, I actually, um, you know, managed to get a few early clients out of, um, just word of mouth and basic sort of just organic network, just discussions about what I was doing, and people knew who, who I was. And so certainly a lot of kind of, uh, our early business has, has been that way. But other parts of the, the sort of the, the more challenging side of things have certainly been to do with, um, personal brand development. So I feel like I'm very much in, uh, in, uh, in the sort of, um, starting phases of, uh, really kind of getting out and publicly sharing my knowledge and, um, establishing myself as a, I suppose a, a, um, a thought leader in the space. That, that's one of the things that I think is really challenging but pays enormous dividends. Um, this conversation, for instance, was the result of me just engaging with a chat on LinkedIn and, um, and you responding, um, which I think is amazing, and I think a lot more people need to realize that. So I think, um, some of the guidance that I might provide is that, you know, I think I sat around on LinkedIn consuming a lot and maybe posting here and there, and I actually think that people have more to say than they think they do a lot of the time. And I certainly was very guilty, I still am, of overthinking a lot of things, and sometimes you just gotta, you know, you don't need to write an essay every time you wanna post something. If something's on your mind, if you've had a great conversation with someone, you know, put it out there. Um, you might get three, three, um, sort of reactions. You might get 50, but it doesn't really matter. I think it's much more about the compounding impact of, um, actually just being seen. Um, so that would be sort of one big difference. Um, in terms of, um, sort of, um, financial sort of rewards, et cetera, um, very much kind of level with where I was previously, but I think that the potential upside, um, and I mean, bearing in mind I'm just a few months in, um, that's been gratifying. Um, but certainly the kind of, um, the need to balance pipeline development with, uh, existing client work is, is a really challenging kind of balance to, uh, get used to. Um, you know, you can't drop the ball with, uh, existing clients, but equally you've gotta make sure that y- you've got, uh, new things coming in. So, so that's, that's been a-- That's something that you don't face if you, if you're working for a company where you've got a guaranteed salary at the, at the end of the month. Um, and then the other side of it, like the, the final thing for me has really been about lifestyle, uh, flexibility. I would say at the moment I'm working much longer hours but much more flexibly than I was before. Um, and I'm, I'm planning to, uh, to certainly sort of dial things down as I start to kind of firm up a lot of the systems required to be able to operate the business much more efficiently. Um, but yeah, I've got three young kids, um, and it's been amazing for me to be able to say, "Well, you know, I'm just gonna be there that afternoon," or whatever the case might be. Um, so from that point of view, you know, it's been, it's been really amazing. Um, one last thing is really about variety. Uh, I've loved the fact that I am now just talking to, meeting, like just so many different people who are faced with this. Common themes in some of the problems, but also I'm dealing with so many sort of different, um, sort of more, um, I suppose like a, a broader variety of different things, which is great.
- AGAakash Gupta
Really resonates with my experience. Yeah.
- ACAndy Carroll
Really?
- AGAakash Gupta
The first year of going independent, I was probably working way more [chuckles] than I used to work. Now that I've figured out some of the systems, it's better, but that doesn't mean that selling podcast ads isn't a slog that I need to keep doing, doing all this work. You know, there's so many different things that go on with creating your own business. But this has been an awesome introduction into Vibe Coding with Windsurf. Andy, thank you so much.
- ACAndy Carroll
It's an absolute pleasure. It's been really good to be here. Thank you, Aakash. Uh, we'll be in touch.
- AGAakash Gupta
All right. Bye, everyone.
- ACAndy Carroll
Bye, everyone.
- AGAakash Gupta
I really hope you guys enjoyed that episode. It would mean a ton to me and the team if you could please subscribe on YouTube, follow on Apple and Spotify podcasts, and leave a rating and review. Those ratings and reviews really help grow the show and help other people discover the show, and they help fund the production so that we can do bigger and better productions. Can't wait to share the next episode with you. Until then, see you later.
Episode duration: 1:12:14
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Transcript of episode oLmHdymHHg0
