Thank you for your comment; I really appreciate it.

I’ve only just started working on this, but I’d like to get some feedback from the community so that it can be as community-driven as possible. As for staffing, it’s possible that I might need to hire a few developers to help me out.

Regarding AI, in the first phase of the project, I’ll likely rely heavily on AI for the artistic aspects (e.g., icons, background), but I prefer local models over hosted ones for the same reason you mentioned. For the code, I’ve been a developer for years before AI even existed, so I use it as a companion to speed up my work on the smaller tasks.

Finally, regarding community voting, votes will obviously be weighted, and in any case, voters must provide a valid reason to vote, so I expect to filter out trolls and the like right from the start. We’ll definitely have to see how things play out based on what develops.

Thanks for the comment and for sharing the link… BEEP BOOP. :D

Genuinely appreciate this, thank you.

You’re right that Lemmy is new territory for me still learning the culture and clearly stepped on some landmines along the way.

On LLMs I’m pretty firm in my position: useful only when you already know what you’re doing, only to move faster, and absolutely not a replacement for understanding your own work. And they get it wrong constantly even then which is exactly why you need to be able to read, write, build and debug without them first.

Local models I’m fully on board with in principle. The environmental point is well taken. The problem I keep running into is that for actual coding tasks, the local options that are genuinely good enough still want a GPU setup that costs more than a full datacenter expecially now with the RAM shortage. If you have any recommendations on that front though models, setups, anything that punches above its weight I’m all ears. Seriously.

Genuine question, not rhetorical: is a rough, unpolished pitch actually more respectable than a structured one, just because no AI touched it? I’m honestly curious about the reasoning there.

My take is that the ideas the design decisions, the systems all of that came from my head, not a prompt. AI helped me organize and present them clearly. If the same thoughts were written in half-broken sentences with no structure, would the content be more trustworthy? Or just harder to read?

Also worth noting every reply in this thread, including this one, is me. No AI, just typing here. If the concern is about respect for the reader’s time, I’d argue that trying to engage genuinely with every comment counts for something.

I do hear the broader point thougth. AI-generated content has burned a lot of people and the distrust is completely earned at this point. I’m not dismissing that. I just don’t think using a tool to structure your own ideas is the same thing as having a machine think for you but I get why the line feels blurry from the outside.

Yeah, totally fair right now it’s just words on a page, I get it.

But that’s kind of exactly why I’m posting here instead of disappearing into a basement for two years and reappearing with a finished game nobody asked for. I’d rather show the concept early, collect real feedback from people who actually play these kinds of games, and build something that reflects what the community wants not just what sounds good in my own head.

So if you’ve got thoughts on what’s missing or what doesn’t work, genuinely interested to hear it.

The game won’t be AI slop, i’ve covered that pretty thoroughly in a comment above if you want the full breakdown. But hey, thanks for comment by either way!

Haha yeah, the name. I noticed the… problem… somewhere between brainstorming and actually building everything out. At that point you’re already attached to it and renaming feels like surgery. Not ruling out a change down the road though, fair point.

On the P2W thing I hear you, but I genuinely don’t think “stuff costs money” automatically means you have to sell power. Riot built an absolute empire off cosmetics alone. League, Valorant billions in revenue, zero mechanical advantage sold. Is it easy to pull off? Hell no. Does it require a much stronger design discipline and a game people actually want to spend time in? Yes. But it’s been proven to work at massive scale. My bet is that players in 2025 are more willing to pay for identity and cosmetics than most developers give them credit for. I’d rather build something worth paying for than sell shortcuts.

As for the website you’re completely right that it’s bare, and I’m not going to pretend otherwise. But here’s the thing: I made a deliberate call not to fill it with AI-generated art or Figma mockups just to make it look like something it isn’t yet. As a player myself, I’m genuinely tired of seeing gorgeous trailers and promo images that have zero relationship to the actual game. It’s become almost a meme at this point. When I have real gameplay, real screenshots, real footage that’s when I’ll show something. Authentic over polished-but-fake, every time.

So yeah the site is bare because the game isn’t done. That’s kind of the honest version of game development nobody likes to show.

BEEP BOOP BUP - variable not found. Rebooting human mode.

Ok jokes aside, let me actually answer this properly. I’ve been a developer for years before AI existed as a mainstream tool, so the foundation isn’t going anywhere. Do I use it? Absolutely. Do I depend on it to think for me? No. There’s a pretty big difference between using AI as a companion that speeds up your workflow and blindly letting it generate a game for you. One requires you to already know what you’re doing. The other just produces slop.

The reality is: AI is only as useful as the person driving it. If you don’t have a clear vision, strong technical knowledge, and the ability to catch mistakes, you’ll end up with hallucinations dressed up as features and believe me, those are always lurking. You still need to understand every system, every design decision, every line of code. AI doesn’t replace that. It just means I can get from point A to point B faster without sacrificing quality.

Now, the pitch specifically? Yeah, I’ll be fully honest here AI was genuinely a huge help with that. Not because it invented the ideas. Every mechanic, every design pillar, every dependency between specializations that came from months of thinking, designing, and iterating in my head. But translating all of that into something structured, readable, and coherent for people who aren’t inside my brain? That’s where AI earned its keep. It helped me formalize things I was taking for granted and present them in a way that actually makes sense to someone reading it cold.

So yes, AI helped write the pitch. No, AI isn’t building the game. There’s your answer.

On top of the other advice, Pillow is definitely worth considering for this. It’s a solid library that fits this use case perfectly.

You shall not pass… into the world of systemd!

Keep us updated!

You should probably look at this from a few different angles. Is the n8n instance internet-facing? And is it actually reading the email content? If so, it might be vulnerable to various attacks. Can a user manipulate the input in any way?

I’d have to disagree; Proxmox is definitely the sweet spot for a self-hosting setup.

I’d definitely go with Renovate + ArgoCD, or any other GitOps-based tooling.

Well, it has to earn its stripes just like the rest of us in IT! No shortcuts, even for an LLM. :D

Have you had a chance to check the debug mode yet? It might provide some extra details that we could look into. If you can, feel free to share a few log snippets here so we can help you figure it out.