Script💻

The Beginner's Guide to Roblox Scripting book (Vol. 1) available for pre-order! It is restocking sometime mid-October. I rewrote the book to have new images of Roblox studio, updated for 2026, with higher quality paper, so that it's easier for you to read. It's been out of stock for several months now, but you will be able to get it soon! Orders will begin shipping mid-October. I've attached some images of the new version to this post. If you pre-order, you will be guaranteed a copy of the book once it's back in stock. You can now submit a pre-order on our website: scriptsstudios.com/

Vibe coding looks cool on the OUTSIDE- tell the AI what you want and it does it. Bugs? It'll fix it for you. What's on the INSIDE though, is SCARY. - Update after update trying to fix teleport issues that you catch in the funnel - Not knowing why players aren't completing a certain funnel step (and realizing it's because of unsupervised AI code) - You've tested a feature 100 times, but in production, players are having random issue after issue with it. - Stuff that is super easy to fix if you understood the code from the beginning, but is an absolute nightmare to fix months into development with some AI code that no one understands. I've consulted with owners of games that are genuinely fun that I think genuinely would get players. But look deep into the code and you start seeing the issues. Vibe coding means you are accepting the short term win for the long term. The best wins in life are long term. You can use AI. Just make sure you understand what it does. And the best way to understand someone's code is to know how to code yourself.

Game feel can make or break your Roblox game. If your game feels stiff, robotic, or just low quality, players are more likely to leave, making your CCU fall and you losing money. But if your game immerses players into it, makes them feel certain emotions as they're playing, with fluid animations throughout, then they'll stick.. and maybe even.. get addicted to playing. It's so easy to create AI working prototypes, but to nail the execution requires experience, taste, and judgement. That's exactly what I help Roblox developers do. If you've built a game or feature with AI but aren't happy with how it feels, I can help turn it into something polished and professional. DM me "POLISH" and let's talk about your game.

The biggest Roblox update I want is to see my average playtime increase by the hour, rather than waiting 1-2 days to see it. Right now, I make an update and I have to wait 1-2 days to see if it actually increases playtime. Session time, sure, we have live performance, but playtime is more important for the algo. Session time isn’t the same as playtime. An update like this will allow me to iterate on playtime basically 8 times a day rather than 1 time every 1-2 days.

I know a lot of vibe coders don't ever look at the actual code.. but just in case you were wondering, AI slop is... well... slop.

if your game is showing bad stats, just accept that your game in its current form isn't worthy of good stats yet. and that's okay. because it's next form will be even better. you will make it better.

dude, all I do is work. and i love it.

Proof it’s possible

New core belief: If you know what you're doing, you can revive a dead (under 100 CCU) game. I got my game to 1.2K CCU, dropped it down to 40 for a few days with all stats dropping, and now it's climbing 200+

Never blindly trust AI guys

I think of AI as a speed and force multiplier on what you already know. AI amplifies the quality of the person directing it.

If AI can do the thing (scripting/building/modeling/ui/etc.) better and faster than you, then what's the point of learning anything? AI will just do it. My answer to this question is that learning is about building new mental models to see reality. Yes, AI can perform complex calculations, can figure out what to do next, has believable answers to most of your questions, can write code extremely fast and make it work with a single prompt. But, at the end of the day, you have to be able to differentiate between what is a good response and what is a bad response. That is a skill called judgement and it's overwhelmingly important in society today. In fact, judgement is why most Roblox developers never see much success. Because judgment is not about "AI make this game for me ok publish boom 100k ccu" Judgement is about "AI make this game for me using these specific instructions. I want the economy to function like this. I understand that one approach is doing it this way, but doing it that way has long term retention implications, so instead, let's do it this other way. I want you to code the DataStore modules in this way, ensuring that you cache ........ Tell me about your plans on how you'll code this before you write any code so I can review it." And then even after it's done writing all the code, realizing that some parts of the game still need improvement and iterating. Judgement is about knowing when the AI's response is garbage, and when it's good. It's about knowing what about your game needs to improve. And you know the best way to develop judgement? By learning and doing it yourself. The best way to be able to tell if AI's script is bad is by having the experience of already scripting, making mistakes, writing bad code yourself and realizing how to fix it. Everyone has access to AI. Not everyone has the best judgement for the best outcome. The top scripters are able to glance at AI code and realize all the long-term flaws with it, and reprompt it using better instructions (or just rewrite it themselves). So, learning is still worth it.

Make your game so good that it’s impossible for it to become unsuccessful. Even if it requires 20+ updates a day and insane obsession.

Released a game a month ago. Few months in development. Took it to 1.2k ccu peak (so far). Released another game last week. Probably going above 1k ccu soon. Releasing two more shortly today.