Build your own theme park
I built an agent that helps you build Rollercoaster Tycoon-influenced theme parks. You can prompt something like “Build me a cool theme park” and it’ll build a cohesive theme park with multiple worlds and connected with paths and rides. The weird part is that I built it using the same ideas we use to make AI-generated websites follow a company’s design system. I work as an engineer at Magic Patterns, focused on building our Design System Agent, aimed to use your existing brand, components, and conventions instead of producing something that looks like generic AI-generated UI. By default,…
In plain words
Build your own theme park is an AI tool that generates complete theme parks based on text descriptions. Users describe desired roller coasters, rides, attractions, and scenery, and the tool creates a cohesive park layout with multiple themed worlds connected by paths. It's designed for anyone wanting to quickly visualize theme park concepts. The tool uses design system constraints to ensure generated parks maintain visual coherence rather than producing generic AI outputs, similar to how brand design systems guide web generation.
written from the facts on this page · September 2026
From the sources
Create your own theme park in seconds. Describe the roller coasters, rides, attractions and scenery you want, and watch your park come to life.
In the maker’s words, at launch
I built an agent that helps you build Rollercoaster Tycoon-influenced theme parks. You can prompt something like “Build me a cool theme park” and it’ll build a cohesive theme park with multiple worlds and connected with paths and rides. The weird part is that I built it using the same ideas we use to make AI-generated websites follow a company’s design system. I work as an engineer at Magic Patterns, focused on building our Design System Agent, aimed to use your existing brand, components, and conventions instead of producing something that looks like generic AI-generated UI. By default, models tend to converge on similar-looking designs. This is part of that “vibe coded slop” feeling: the same typography treatments, shadows, cards, icons, animations, etc… My job is to figure out the right context and guardrails so that when you prompt “Build me a dashboard,” the result actually looks like your existing product. At some point I realized those same ideas could be applied to RCT. While in web, you might have rules about which typography, colors, spacing, and components should be used together. In a theme park, you need rules like: rollercoasters need complete tracks, rides need entrances connected to paths, paths need to connect different areas of the park, and a pirate-themed world should actually use pirate-themed scenery. It was interesting seeing how closely the problem of building a coherent theme park resembled the problem of building a coherent product UI. Similarly with web design, simply giving the model the right components wasn't enough. I ended up building an eval loop where Magic Patterns would generate a park, another agent would grade it against a rubric, and then the agent would update its rules and skill files before trying again. The rubric checked things like whether rollercoasters formed valid tracks with at least one drop, whether rides were accessible by paths, whether each world used the appropriate themed scenery, and whether the park worked as a whole. After a lot of iterations, you can now prompt something as simple as “Build me a cool theme park.” and you should get back a fully functioning park with guests, rides, paths, rollercoasters, and themed worlds. Best yet, you can also follow the guests, listen to their thoughts, and watch them ride the rides in RCT fashion. Happy to answer any questions, but would love to see the parks you all create!
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, August 2026
the whole month →- TL
Life & fun · 10d ago · louisabraham.github.io
- SA
Hello HN! I found that picking out plausible but diverse skin tones for my digital art and game development projects was kind of difficult, and I got curious about if there was a way to define a color space that made it easy. I've built a color picker and procedural generation algorithm based on the space as well as a bunch of other fun js features and demos throughout the page that use the equations. If you find it interesting, I have lots of explanations of how I built it and what properties the space has. The methodology might be a bit shaky, but hopefully the result is as helpful for…
Life & fun · Aug 2026 · toneyalexander.github.io
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com