Every AI Coder Needs This Roadmap (Myself included)

Let's talk about the vibe coders roadmap or just the product development life cycle in general. I think a lot of people are missing the boat on this. We have a lot of new entrance into the space that did not have time in tech before this and we also have a lot of people that were one part in a larger process that may be missing some steps. So I've spent a lot of time thinking about this working in this space and I think I can be helpful for that. So this is a daily upload channel. We cover a couple of questions and answers. We cover any relevant news and a take on that and then strategy. So we will be covering the vibe coders roadmap in a moment but this actually is relevant. So first of all cursor is putting out some good guides on documentation and how to use it as best as possible within the realm of using cursor. They also put out a guide on using architectural diagrams and this plays into the road map and they also have another one about setting it up for web development. If you aren't already subscribed to my other channel on the Parker X main channel, that channel has three videos a week. There are deeper dives on topics like this. So I'll go through documentation and digest it and make it much much easier to understand through the lens of an actual project that I'm working on. So it's beneficial because I get to go through and I get to implement said thing into one of my projects and then you get to see how it works. The other stuff, this is actually included in the web design or web dev setup that cursor recommends. It's called frame link. So, I'm pretty excited to give this a shot and I'll be applying it to some of the designs that I'm working with for the new Vibe with AI product. So, we have a community and right now it's just on Discord, but also building out user profiles, a thin client on web where you can meet other members, see what they're doing, see their activity, see what their skills are, see where their skill gaps are. And so as I start riffing on this, then I'll go and I'll use something like frame link to bring it into cursor. And that's here. So cool. Next up is questions and answers. Oh, just kidding. How about these sonnet 4 requests? I built all of this weekend. And this is what it looks like. It has this is the VI product that I'm building for our new offer that I'm launching this week. But yeah, pretty awesome that Cloud 4 was able to help me get this stood up. It is an Astro front end, very thin client just for the actual like marketing site and then blogs in the future. And then it will be a React or sorry, Tanstack start or Tanstack router, not sure which one for the application shell. Then it has a Discord client, a Val Discord bot, that's our co-pilot that we're releasing, bunch of other cool stuff. Don't want to get into that in this video, but it follows all the steps I'm going to talk about in this road map. So, next up is questions and answers. We had a bunch of people asking for prompts around Notebook LM. If you have not checked out this video, I cover what I wish I had when I started coding because you can take all of the different video content or blog articles and create a bundle out of that and then break it out to a mind map and it's a huge timesaver. So this one's actually a mind map of how Notebook LM works. So pretty meta, but I go into how you could apply it to other use cases, how to customize it so it's faster and better, how to break it out for the example I use was an analytics API with fast API and time scale DB. Pretty interesting. So a lot of people were asking for that prompt. If I don't reply to your prompt question within two days, you can bet that I'll get to you within three. So, this one guy was like, "Yo, what the hell? What's going on, man?" No, I'm not lying to share or lying for engagement. I just didn't reply that fast. Other questions? Let's see. I'm waiting for the cursor background agent video. Yeah, it's pretty cool to be honest. Hello, Helenios Maluc. I don't know how did I say your name right? I'm sorry if I didn't. But yeah, I've been fiddling around with the background agents just a little bit. I'm also excited this week to be working with Augment on testing out their background agents. So, this one's a little bit buggy, I've noticed. So, if you do command and then semicolon, it'll pop it open, but it actually never ends up loading on the left, which is pretty pretty weird. But I will have a video on background agents and I'll actually add this to our repo. This is open source so you can see what videos are coming and I'll be announcing this repo later today. But cursor background agents video. Cool. And that will be on the main channel. Great. Up next, do we have any more interesting questions? We need a boilerplate bootstrap for AI coding projects. Yeah, I totally agree. I actually announced that I was going to make this Turbo repo or Turbo AI project probably like two and a half months ago. And I started to do it and I just realized that it was going to change so many times that I needed to actually build better infrastructure around all the different projects that I'm working on so that I could open source all of them and have one layer up of templatizing the way that I do open source. So that would be included in our Vibe with AAI repo. So that's not it. But if I go back into here, there's a few that are public right now, but we have the Vibe withAI project which is releasing. We have the pipeline so you can see what's coming, comment on it. We have the AISDLC, our software development life cycle. And this includes seven of the different steps and prompts that you can chain together for the Vive road map that I'm now going to cover. So that's it for today on the questions. Now let's talk about the software development life cycle. The vibe coders road map. So the title of video is the vibe coders road map. Really it's the software development life cycle. So if you typed in like software development life cycle, this is a topic that I went in depth on in our community, we do a workshop every two weeks where I basically take all of my learnings on a certain topic and go deep on it. It ends up being about a 20-minute talk. And this case, here's like a basic version of this, but it's like discovery, design, development, testing and QA, release, then maintenance. There's a lot of different ways. Like this is super basic, but the way that I think about it and how it maps to the tools that we use is in this session I basically talked about PRDS and the importance of them. And that's just one part of the puzzle. So if you break it out even more, you get whoops to where yes, you do need to have crisp product requirements. Why should you listen to me? Because I've done a lot of stuff. Here's me running next to a bridge. But you want to get to a point in your life cycle where if you're going to be leveraging AI, which you should, if you use more detail, you'll get better outputs. Garbage in, garbage out. So if you compare the thing on the left, this is something done with the correct prompting and the correct planning. This is on the right with Taskmaster. It's not a hate on Taskmaster. I think Taskmaster is great for certain tasks, just not the complex ones. And so again, you can't skip out on the thinking. And something to note is that the quality of the product or the quality of the outputs from AI goes very down very fast. It degrades very quickly as you do multiple turns. So it's on you to make sure that the put like the things that you're putting in are the highest quality possible. And sure the models will get better, but you want to be ahead, right? And you want to get those skills because those skills in my head I think these are the most important ones. Systems thinking, systems architecture, pattern matching, creative problem solving, debugging, and ideiation. Those will always have a premium. And so you want to spend time on planning. Let's get up here to the actual process. So from a high level, this is what it looks like where you have task generation at the bottom. But again, we're gonna want to go a layer deeper where I was doing some research and I think that it depends on where you are. Like I'm assuming that this audience is people that we have like three buckets. We have business owners who understand that AI is going to be helpful and that they want to dip their toe into it and start using it. So they're brand new, but they're very incentivized to get something working. The second bucket of people in vibe with AI and the viewers here are people that work in tech or are adjacent to development and they see it and they want to use it in their job to get better. Maybe they have a nights and weekends project. And the third are solo founders or full-time entrepreneurs. So, I was going through this and I was doing some research on all the different parts of the puzzle because I'm trying to get better at this topic. And this topic, this is pulled from eraser.io. Oops. And eraser.io. If I just can finally get the window pulled correctly. It allows you to do all the different kind of diagramming for both your cloud resources but also the inputs and outputs of databases like figuring out what the systems architecture of your software is going to do. So there's different layers to this and I'll be frank, I'm trying to round off the corners and sharpen the edges. Those are opposites. I'm trying to get better at the different areas of this flow. So I had taken this and I'm trying to build out a system that includes both my business goals and that flywheel and then also the systems architecture for my software. So I've been thinking a whole lot about this. I can show those towards the end. But when it comes to the different steps I started there and I said what would you consider this and this could be data flow architecture a process flow component integration. So those three are you know how the information moves through the system from the scraper to the database to the various processes. In this case this is something that I'm thinking a lot about which is how do I automate the grunt work around YouTube videos before the video is shot and then how does that play out after the video is shot and so that's a product I have called Echo. And then how does that play into the grander scheme of hitting 100,000 subscribers, hitting a certain number of Discord members, and also increasing the quality. So I believe with the right automations and the right flows with your road map where you're dialing in the pieces that for the first time ever, you can actually have something be done faster, better, and cheaper. Now, we'll see how that plays out. You'll notice that I'm spending a whole lot of time on this system stuff. So then you have process flows. So that's the content workflow. I'm going to be putting process flows next to the data flow architecture because the business in the process flow drives the data flow. If you don't have the business requirements, if you don't know what it is that you want as your ideal 11star Taj Mahal state of how you want to operate day-to-day, then how are you going to build the product to support that? You can only get the process flow if you've done it manually a bunch yourself, right? So that's why you've seen me do these videos. I've been making videos for a little over two months now and I've done most of the parts manually and I'm just learning which parts of the process need to exist, which don't. Then you have component integration, how different services and tools connect together. So this is all just one part. This section is just one part of the VI code roadmap which really is a misnomer because it's really just software development with AI. Next, you actually get a system design or architecture planning or technical documents or solution architecture. And these are the topics that you'd study. So, I would use a prompt like this and say, how do I get better at this? What is my learning path? What are my tools to practice with? So, that's when I see Excaliraw. This one does not include eraser, but eraser is really helpful. I've actually found myself using it a bunch. And so then I say, okay, great. Break out a list of the topics I need to upskill on within this list. Con context, I'm a skilled engineer across the stack, including some DevOps. I need to use these skills. It goes into reasons. Then it breaks that down even further. So again, we're going deep on one part of this road map, and I'll show you the whole thing in a second, but it's just to highlight that look, there's a lot of stuff that's involved in this if you want to build a bunch of software that is high impact that can run at scale. That's a phrase that I probably shouldn't use, but this is how you can know the full gambit and you can pick off pieces that you want to focus on. But let's get down past this. So I got this kind of study guide. Here's the critical path that I had because I said look I want to Pareto this understanding to the 8020 of what I need to learn based on what I'm building. And then I pasted in a summary of what I'm building because we want to do just in time learning. So I have had trouble building out a lot of these systems because I'm weak in this area. And that's part of it is just being honest about what are you weak in? What are you good at? And so then you keep going down. Here's some examples of orchestration. And then finally, I pass it this, which is okay, cool. I need to do everything manually now. I'm trying to build out the system. What is the standard software development life cycle? Previous product manager, then software designer, then software engineer as the last three years. This is my understanding of it. And then I passed in what I think to be true. And it actually gave me a really awesome set of things to work on. So the problem definition is identifying real user problems. This is the first step and this is what I cover when it comes to the first step in the AI SDLC which is now available. You can check that out. I will put the link in the description. Actually, I will. So, let me put that as a to-do video description. Add the SDLC link to video description. make video about SDLC that way people understand how to use it. So first is the problem definition and this breaks down a couple things I have for each one of these sections is what it covers the role in the ecosystem. So that's how does this play into the larger picture? What happens without it? So why you should care about it? What are the tools you can use? What are the job titles that map to this process at a company? And then some resources. The mom test book is actually really good. Highly recommend that. It is great for user interviews, understanding things. I really like building products for myself because the ones that have been successful are the ones that I know the problems very deeply. And then you don't have to loop back and cycle and cycle and go and talk to people in order to get your first version because you can just dog through the whole thing. The next is the product requirements and user stories. So this is part of the F1 prompt that I always use. This is from the AISDLC 2. So you don't actually need to comment in order for me to paste this back. You can just go and start that repo. But that finds all the open questions and creates user stories for you. Next is the UI, UX, design and prototyping. So, this is one step I feel like a lot of people are missing is they don't actually go and draw stuff. They just try to prompt their way around and vo. And I still think if you're bad at Figma, get decent at it. There's every Apple awardwinning, not every, but there's a ton of Apple awardwinning apps that are redesigned in there. And the best way to learn is actually just go and see how they're made. You don't have to draw all the shapes yourself, but you'll start to notice the nuances and the subtleties of design. Where are the shadows? How do they deal with these little subtle borders? How do they space things out? How do they work from mobile to tablet to desktop? What do they show? What do they hide? You basically just don't want people to think. You want them like my best recommendation is to boil down all the functionality to solve their problem as fast as possible. You want to reduce the friction. So that'd be the third step. Fourth is the technical architecture and system design. And what I recommend as you're going through these is just give yourself a grade 1 through 10. If you don't know much about it, then you can't even grade yourself. Just throw it at zero and then make that a practice where every day for 20 minutes you are spending time on this topic going deep on it. For example, I'm now looking at this, which is a guy who just teaches you how to do the systems design interview because it's helping me better understand. Oh, cool. How would you design YouTube? And you have all these examples of it building out the system. And then you I read one of these and I'm like, I don't even know how transcoded storage works. So, I have a this one open and then I have a little GPT window open. I'm asking a bunch of questions. And this covers the database schema. I know that I know that well the API design pretty good at technology stack selection getting better at that. Scalability plan security considerations all this stuff. So then you get architecture decision records. Didn't even know that was a thing until I got this. And so then you end up with some resources which is oh designing data inensive applications. I have a folder full of these books that I can then just give to a GPT ask a bunch of questions about. Then you have database design and data modeling. That's actually doing the schema relationships yada and all the different tools. As you can see, there's a lot of stuff in here. Then you get into backend development and API design, front end development, user interface, off and security. I thought it's interesting how I've always just done off the get-go. So I think for this is like traditional step by step. I'll probably switch this around a little bit because in my head I want to have all the basics done when I start a new project and especially if you're doing this is 0ero to one to 1.n N would look a little bit different. Then you have QA. So I would think in my head these map to agents or agentic workflows where it can go and do it or I can just train a folder full of good examples with a prompt because as these models get better, they'll be able to have more context. Then you have DevOps. I feel like a lot of people miss out on this one. I know that I sure did because if you're just building simple web apps you can get away with versel but no that's not what we want and then you have CI/CD all those topics performance optimization document and knowledge management user feedback so then I say all right as a solo founder which ones should I prioritize and then it pairs it down to these so that is the road map and I'm happy to share this as a PDF I can put that somewhere. But I think it's important. I think anyone that watched this video, I know from me just talking about it that there's a lot to it and you just have to prioritize which parts are you in need of the most for the thing that you're building and then prioritize them and go and do the homework, figure stuff out, get a baseline understanding of it and immediately get to building. So I don't say go and just study it for 20 minutes a day and not do anything. It's the 20 minutes of reading that then encourages you and gives you the confidence to go and actually implement this stuff. That is where I'm going with this. If you learned anything in this video, make sure you like it. If you want more videos like this is a daily upload, so you can see one tomorrow and the next day and the next day. And I would love if you checked out the community. We are launching our new product which I'm very excited about. I can give a little teaser of that now. It's got a co-pilot and a lot of people are saying nice stuff. So, let's see. The first version of it was we did this paid educational content. Next is this builder network. So, we're really tightening up the ICP from being wide audience because I was learning like who's the customer, what type of person is this resonating with to okay, these are people that want to build. And we do have those three cohorts that I spoke about, but we all are tied under the same mission of wow, we can leverage AI to build everything and market it. That's the other side of it. So it'll include the content builder network, agents for learning, dev tools, agents for distribution, and repo access. And then why does it matter? You can meet the serious builders that are leveraging AI. You can decide if you want to work with some people in there and pair with them. You can sift through the noise with the news agents, all that stuff. But the one I wanted to highlight was the upgrade to Discord. Val, that's S. Val has a set of co-pilot commands. And so you can essentially get the gist of what's been going on in Discord with one command. So you can proactively do it and say, "Hey, slummarize," and then select the duration that you want, and we're indexing all the messages for that period. Everyone in the Discord when this launches will be aware of which channels those are. They're mainly like the tool ones and the AI news ones because they're really noisy and things change. And so that will happen. Then you also have proactive ones which is on a cron and it'll run that against the vector endpoint calling the index and we're using Gemini the Google stack for all of this. Also have TLDDR which is pretty cool. So, if someone drops in like a 4-hour video and you just literally want the bullets, you can do that. Coming soon, we'll have birdie, which is a news co-pilot. So, before on this channel, I would do like a whole bunch of news. But now, I just want to do takes on certain stories. And this will just run in the background on the Discord. You'll have different roles that are associated with which ones you want to have notifications on so it's not noisy. I recommend just putting on notifications for mentions, but that's a different story. And then there's other things in here, but yeah, I'm really excited about this. This is what I made over the weekend using this stack. So, a shout out to Michael. He is going to be helping. He's a member first of all, and he's going to be helping with the application shell. And I'm just excited about this because if people want to work together on stuff like in the case of I'll use Michael for an example. If we decided, wow, this is really helpful for our community. I'm sure other communities would want these co-pilots and this kind of member management portal. Then we can decide, oh, let's go sassify this thing and figure out what steps those take. So, pretty pumped on that. There's a whole bunch of other stuff, but yeah, that's for the members to hear about later today. I'll catch you tomorrow. See you.