I've seen some cool enterprise AI dev tools. There are some very smart people working on these problems out there of course, and they're all working with AI too so the solutions are coming fast. While it's interesting to me - and their stuff looks pretty awesome, it still looks to me like they are trying to put AI into the existing SDLC (Software Development Life Cycle), I guess they kind of say as much. Which yes - has its place for sure I guess. In a big enterprise maybe. And they speak with fluent jargonese - which for developers that makes sense. Their sites are complicated and business-y. But the world of development has opened up to everyone now. That's why I expanded SDLC into what it stands for above - because you don't need all this jargon and geek-speak to build software today - your agent can do all that. But it needs to speak your language. We all are developing now.
But to me - I think to my cofounder and I as we started this - it was always about "throw out everything about how we used to build software. And start with a clean slate - see what a product person with no dev background does today, and make that work and scale."
That's where CatWrangler came from. A dude/dudette sitting there with Claude, telling it what to build, and yelling at it when it screwed up, and fixing and changing and moving on step by step - one decision at a time- until you had something built. That's the most efficient process if you're still going to involve humans. Not spending all this time with a human (and even ai help) writing detailed specs for what is going to be built, not breaking down every feature idea into an SDD and then different chunks of work to be parceled out piece by piece to different people/ai's. It was just "hey, let's build this," and then "Hey, let's add this feature" and seeing where it went. And when you added that, suddenly you needed THIS, and then you also needed THAT, and then the AI suggested this too, etc etc. and you are just moving - one idea at a time - one decision at a time - building this pile of choices that are now part of your DNA, they are where you came from, and why you did what you did. These ARE your product. And while you're in the moment, with your ai, you are thinking deeply about each of these problems. You hold the model in your head, of that specific module, that specific decision - and you get it right, you feel good about it, and you move on. and you will NOT remember it a day later. Even hours later you may not remember it, or certainly not everything about all the contours of the decision. And your ai will not either, because it will have reset 3 times, 10 times, whatever - since then. But now we have captured all of that - and stored that in the system tied to the code. So every ai who hits the system now knows - and can walk this decision tree to build your app. And you can too if you want to. Now we know there was logic behind what we did, and what that logic was.
I want to capture these posts directly from my thoughts- and these are unedited by ai (unless I say differently) because we are surrounded by ai, and it's all we're doing, so I thought that thoughts from a human, directly and unedited, might shed a slightly different light on all of this. Now that being said, our first entry was written by Claude, by an ai - first as a placeholder but it was so good I told it to leave it! And how appropriate - because AI's are really our customer. You are the product brain - you want something done, and you have an idea for how to produce a digital product to do that thing. You don't care really how it's done - heck, a week after we had started building this Micah asked me what language it was written in, and I laughed, because I didn't know. Neither did he. We didn't care. Welcome to the new world of software development.