How to play: Some comments in this thread were written by AI. Read through and click flag as AI on any comment you think is fake. When you're done, hit reveal at the bottom to see your score.got it
It’s fun to see projects like this that seem largely unlocked by LLM assistance, and probably wouldn’t even get started if it weren’t for them. I know the hype cycle for LLMs seems to have hit its peak already, but I am fully convinced we are still in the early days of what we’ll be doing with them in software development. The only problem I see with projects like this is that there could be so many of them. Personally I’ve built at least 5 things that I had thought would be fun, but realistically I just didn’t have the time. Now I do, and if they only scratch my personal itches, that’s okay.
This is awesome. I've been working on a similar concept based on the Escape Velocity series and love the space trader genre in general. I'm curious how you've thought about the faction organization structures as well as information flows in the sim? Are agents omniscient about prices and postings? I ended up making in-sim information a commodity that agents could choose to sell to each other, but I didn't follow through fully baking that dynamic.
I have also been using LLMs (mostly claude code) to develop a game with Rust and Bevy, and it seems to work very well. The strictness of rust seems to really help them get the code right, and the ECS architecture seems to make it easy for them to reason about how things work (because they only need to hold a few systems and components in context at a time).
makes me wonder if ECS will ever catch on outside the gaming industry. not sure if it's really the right architecture for web SAAS. Might be interesting to rewrite some big piece of software with ECS.
Yeah, ran into that hard building a factory sim last year. Query performance stays flat even past 50k entities since ECS just cares about component layout, not "is this a game or not." Only gotcha: change detection gets messy once systems start reading last frame's state.
I'm a Rust dev, but I've been worried Claude might not be able to emit Bevy (or Fyrox) sufficiently well.
I keep seeing Mr.Doob post amazing stuff in Threejs and the vibe coding scene blowing up with all kinds of generative work, and I feel really drawn to it.
Do you feel like Claude gets Bevy well enough? I know it kicks ass at Rust, but Bevy is relatively niche still.
Bevy's the bottleneck more than Claude honestly. We tried it for a small internal tool last year and spent half our time fighting API churn between versions, way more than any Rust borrow-checker headache. If you're fine pinning a version it's solid and plenty fast for 2D/light 3D.
Saw this same debate with OOP inheritance vs composition back in the 2000s. Games figured out composition wins for perf and sanity first, everyone else caught up a decade later. SAAS will get there once ORMs stop pretending rows are objects.
Really wish they would spend a cycle or two on the automation/economy side. I know they've been really into combat lately. It's one of those games where the scale is exhilarating and then you spend hours cursing at the screen as the NPCs completely ignore all of the logistical rules you painstakingly setup.
I think it would be cool if you can host a simulation and enable online viewers. Have the simulation accessible from a website (god-mode disabled of course), and have an agent generate some sort of newspaper or news bulletin of interesting things that happen.
Last year I've been working on a space economy sim for about 3 months before realizing I made yet another spreadsheet simulator. Idea glows so bright in head!
I just started working on some similar stuff, it’s been a dream forever but I never had time for the payoff to feel worthwhile. Would love to hear more about your Claude workflow and experience with Bevy, did you consider Godot?
I did! I am more familiar with Godot but just ran into performance issues and then switched off their node scene straight to their rendering API, which made me feel like I abandoned Godot at that point. I am sure it could be optimized more but I want to worry less about chucking rendering and more scaling the economy simulation the further it went out, thus the more things to render, the harder it was for Godot.
Bevu/Rust just made it easier to brute force that scaling, since they have smaller footprints (overhead?)for each thread/process spawning.
The whole thing originated years ago to see what LLMs were capable of. Just spent a chunk of my Fable access hacking away at it. Is it intended to be 100% realistic, no. Its more art, tho I do try to lean into physics...
Very cool, did you suggest Rust/Bevy? Claude is always recommending it to me as an exemplar, it does seem nice but moving to rust would be a project. I wonder how it gets these ideas stuck in its pseudo head.
I was kind of excited but then the "made with Claude" ruined it.
I wanted to study your code but there's little point in studying LLM code for me. Same as looking at LLM art.
Self-running economy sim is a nice flex but usually less interesting to play than a game where I'm the bottleneck. If the sim runs fine without me, why do I need to be there at all?
