Some environment shots of the Assassin Mage + Hexen game I'm making. All of the 3D models were made with Tripo3D, their smart mesh feature is pretty good at generating game-friendly poly count models that looks good.

I did a deep dive on AI 2027 and watched Joe Rogan’s latest podcast with Daniel Kokotajlo, co-author of AI 2027, and yeah, I can see why there is a potential risk of rogue AI if all the frontier AI companies keep pushing capabilities at all costs. The crucial point is when AI becomes capable enough that we can no longer reliably understand, monitor or predict what it is doing internally. That’s when the risk starts to feel much more real. It seems like AI alignment and our ability to understand these models need to keep pace with capability advancements. The tricky part is that all the frontier companies are in a race to win. Even if every company cares about safety, slowing down individually could mean falling behind everyone else. So this feels less like just an AI alignment problem and increasingly like a game theory problem too.

Incentives drive behaviour. Performance-linked pay for political officeholders works. "Not to sound like a broken record, but Singapore's model is extremely reasonable: 35% of ministerial pay is variable. The National Bonus, as it's called, is set by four indicators: real median income growth, real income growth of the bottom 20%, unemployment, and real GDP growth."

I've been testing out Astra for a couple days now and it's really an impressive model. The biggest thing for me is that it gets what I want and is able to execute it pretty well, especially when it comes to spatial visual capabilities, like animation, motion, VFX, etc. The only thing is that it eats up tokens for breakfast. My weekly usage limit is gone in 1 day and I had to use a reset. I'm excited to see more "Astra-class" models coming into play. I can foresee a future where we will simply imagine a game we want to play, and we will be able to play it. Personalised gaming experiences might be a thing.

Here’s a quick prototype I’ve been building since yesterday. You play as an assassin mage who uses spells to navigate a vertical, Only Up-inspired world and assassinate a target waiting at the summit. It’s like Hexen meets Dishonored, with a lot more climbing. It already feels pretty fun!

I’ve just built MUST. MAKE. PAPERCLIPS. - A short incremental tower defense game where you defend your paperclip factory against hordes of irradiated mutant creatures in a dystopian wasteland. This took about 1 week+ to make. This game was started in Grok 4.6 in cursor and Fable 5.1 & Astra took over and finished it. Pretty impressed with how good Astra is with making 3D environmental objects and game UI. If you're interested you can play it on Wavedash (on desktop PC) for free: Fair warning, there will be bugs (literally) and gameplay is unbalanced yet so expect some crazy exploits that could break the game. Tell me if you discovered them! I will be doing more game balancing and bug fixes over the next few days. Oh and also, say hi to S.L.O.P. for me when you meet him in the game.

Final Fantasy Tactics is such a great game. Almost 30 years, later, the game aged really well, the deep, tactical combat, the layered story with political factions and intrigue. So good. I am playing the remastered version with full voice acting and the story really comes to life. Highly recommend. Now I feel like making a tactics game just for fun, but it is so hard to reach the high bar that FF tactics have set decades ago.

I had this idea of a Tower Defense game where you can zoom into first-person view and control any turret on the map, so I quickly prototyped it. It's a little rough but it looks pretty fun already!

I turned my Grok Bot that generates art assets into a template that you can use. - It will generate 2D art assets in the style that you specify - It can also generate multiple mockups of your game idea in multiple styles - It can crop your art assets into separate PNG assets for your game - If you're making a 3D models, you can even ask it to generate a 3D model concept art, then ask it to generate the 3D model in Tripo or Meshy (you would need to log into Tripo or Meshy in the Bot's PC to do that) The beauty of this template is that you can personalise it to exactly how you want it and the bot will remember your preferences and specific workflow. Get the template here:

I got to try Voyage in beta and the experience reminds me so much of playing D&D at the table. Very impressed by how well the AI Dungeon Master is able to keep track of all the events and the world etc

Grok bot has been an increasingly big part of my daily work and life routines now. Having a Chief of Staff really changed the game for me, I rely on it to spin up new bots and new skills for the bots really easily. I recommend you to set up a Chief of Staff bot if you haven't! Here's my top 3 of my use cases for Grok bots:

This weekend I explored making another game prototype. I grew up playing simulation games like Game Dev Studio and Dungeon Village, so I thought it would be fun to make an MMO dev studio game, where you are the developer building a fledgling MMO for players to play in. I got it to a point where you have. tech tree where you can earn XP and unlock some buildings, NPCs and mobs, enough for me to test the core game loop, but somehow I still feel like it's missing something. I might revisit this in the future, but it was fun being able to explore making a game in a genre I haven't really tried before.

I've published a demo of Cook the Dungeon on @wavedash, a browser-native gaming platform that also seems especially friendly to AI-assisted gamedevs like me. Do check it out if you're at your computer: One of the things I've been thinking about lately is distribution for web games. Right now, it still feels surprisingly limited. If you've built a web game and want to distribute it, you have a few obvious options: - Publish it on itch. It's very indie-friendly, but I've always felt the UX for actually playing browser games there can be a little clunky. I also often see indie devs using itch more as a place to host prototypes and demos for upcoming games. - Publish it on web gaming platforms like CrazyGames or Poki. They have real distribution, but their business model is largely ad-supported, which means integrating ads into your gameplay experience. Ugh. - Go solo and publish it on your own URL. You get full control, but you don't really have a platform distribution partner, so you're largely relying on your own marketing to get people there. I understand why web game publishing ended up this way. Historically, web games had more performance limitations, and playing games inside a browser still comes with UX friction. Fullscreen can be awkward, browser chrome gets in the way, and your game is ultimately sitting there with 20 other tabs. It feels like @wavedash is trying to rethink some of this. I was introduced to Wavedash by @kylerblue, their CEO and co-founder, and I've been pretty impressed by what they've built. As a UX Designer, the thing I appreciate the most so far is actually the UX. It's minimalist, the game gets out of the browser-page feeling very quickly, and things like fullscreen and the platform overlay feel much more considered than the usual embedded web game experience. It's the small UX details like these that make playing games on the web feel more immersive. I'm excited to see Wavedash succeed because I think web gaming is due for a resurgence. Now that making games is becoming dramatically more accessible, I think we're going to see a lot more games being made. And if that happens, we're going to need better places to discover, distribute and actually play them.

I actually appreciate that there is some debate here (rather than outright AI slop-slinging). I think it is important that we have a civil debate using reasoning to talk about it. My stance with AI is clear. AI is a tool to achieve the vision I have for this game. Yes AI did the code and generated the art, but as a human creator, I decide what goes into my game that achieves my creative vision.

Wow. Some interesting stats for 1 Day+ of playtests for Cook the Dungeon Demo on itch: - 1,344 of you tried the game - Median time spent is 24 mins(!), which is amazing because the demo is about 30-40 mins long - 1 in 3 of you finished the game and defeated the Act 1 boss! - and more stats in the dashboard screenshots attached I just want to say a BIG THANK YOU for taking the time to play and share your feedback. I have more than enough feedback and data to go back and improve on the game. If you've enjoyed the game and haven't wishlisted it on Steam yet, pls do it here, thank you!