MindHash

Our fully autonomous QA Agent is running tirelessly to find issues in SPARQ that our other tens of thousands of unit, integration and end-to-end tests couldn't catch yet. It can use the engine essentially like a regular creator and automatically report bugs to us. These bug reports are then picked up by our devs or another automated bug-fixing agent. 🤯

we don't blame anyone for doubting us, this industry has given people plenty of reasons to be skeptical that's why we'd rather let what's coming speak for itself

@neutanent this is the part nobody can vibe-code their way into anyone can demo a renderer. running matchmaking, auth, asset pipelines, crash ingestion and 50+ services across regions with auto-scaling is a different sport entirely.

"Mi mi mi why don't you just launch already?" Well, apart from the already insanely complex game engine itself, we literally have 50+ backend services, dashboards, control systems and more that we are running for SPARQ. This is not a quick deploy, it's an insanely complex, multi-region, production cluster with redundancy and security systems, as well as massive auto-scaling etc. Here are just a few services that are not security sensitive that I can share. The production setup is now complete, but yea.. The $NEURAL team isn't fucking around with this one.

This is what it actually takes to operate a roblox-like gaming platform. Mindblowing how much work went into this 🤯

@neutanent "just carve it out in seconds" is doing a lot of work here anyone who's tried building destructible terrain in unity or unreal knows how insane that sentence actually is $NEURAL 💪

Just recorded a quick video to show the Voxel-based Terrain in Sparq, which will allow you (or your agent) to create multi-layered, destructible Terrain out of the box ✌️ You can literally build as many terrain layers as you want in it, seamlessly, and without awkward 3d mesh workarounds. Same goes for caves and tunnels, literally just carve them out in seconds and then paint them differently. Available directly in the @SPARQWORLDS Beta on Steam and will soon help elevate $NEURAL games and worlds to a whole new level 💪

Neuro is on it nonstop. I speak to him daily. As a builder myself, people dont know about the parallel and what it takes. Notice the ones that build in toughest consitions. $NEURAL has great vision

Ngl, this week is quite a rollercoaster of emotions. On one hand our production cluster for SPARQ is ready, and we are doing final tests before launching our beta on Steam. 1 1/2 years of work and now it's finally about to get into the hands of the creators we built it for. On the other hand, $NEURAL has been hit unexpectedly by an old KOL exiting all of his ETH token holdings, including all hus 1.3% of the supply of $NEURAL within 24h. In low volume market situations, this is obviously less than ideal and it came very unexpected. Nonetheless, this doesn't change what we are working towards; Redifining what it means to create games, and to experience them. Creating new markets - in this overlap of crypto, AI and gaming - people can't even imagine yet. No matter what, we will keep pushing til we achieved this goal 🤝

Some reflections today on the complexity of building a game engine compared to other types of software out there. You see, every other kind of software is allowed to be late. Your bank app takes an extra second and you wait. A spreadsheet recalculates and you sip your coffee. A game gets 16 milliseconds. That’s it. Miss the deadline and thousands of people feel it. Here is the rule underneath: a game engine is two clocks. Rendering runs as fast as your hardware allows. The simulation runs on a metronome, fixed ticks, same size, every time. Most engine craft is hiding the seam between those two clocks so well that the player never thinks about their existence. The industry keeps re-learning what happens when you skip this. Dark Souls 2 on PC drained weapon durability twice as fast at 60 frames per second because someone counted frames instead of seconds. Quake players spent two decades locking their framerate to 125 because a rounding quirk let us jump higher. Old PCs shipped a Turbo button that existed mostly to slow games DOWN to the speed of the chip they were written for. None of this is trivia, rather it is the difference between a demo and a game. A demo has to look right once. A game has to behave identically for every player, every frame, every machine, for years. People are out here arguing about whether AI can generate content. Wrong layer. Generation is not the hard part. The hard part is that everything generated has to run inside the tick, on the metronome, without breaking determinism for the player on a five-year-old laptop. That are the parts we obsess over at @SPARQWORLDS... the invisible machinery that makes generated code safe to run. Players will never see it. That IS the whole point. playsparq.com

we spent months questioning every part of game creation because we knew there had to be a better way after months of research, we found our best approach to redefining game creation

@neutanent @SPARQWORLDS

There are only a few steps left for @SPARQWORLDS to go into beta on Steam next week: 1. Finish setting up the production cluster (almost done) 2. Merge in a few more updates (in progress) 3. Test everything inside out (again) 4. Launch Some highlights we are launching with: - Insane AI Agent support (via MCP or in-engine agent) - Real-time multiplayer editing - Destructible voxel-terrain - Asset and character library - Fully customizable editor - Our own package manager (SPM) And so much mooooore. Purpose-built for a new generation of game creators, coming very soon 🫡

how is the financial situation now in the project? wth access to capital on the web2 side, revenue generated during the good times, and a disciplined approach to spending, we've put ourselves in a strong financial position while continuing to build we don't see any concerns about running out of money

Claude has been the reason we could pull of SPARQ in the time and manner we did, but I'm honestly so glad GPT 5.6 lives up to the hype. It feels like @AnthropicAI hates their power users. We have easily consumed 200B+ tokens on Claude and are - objectively - building one of the most complex and interesting products to date on it. But the past 1-2 months have been an absolute nightmare. All our accounts getting banned for no reason at all, multiple times and by some automated security flagging system, no way to appeal, absolute intransparent offerings and a general "fuck you for using our product" attitude are just too much. (No we didnt break any rules) I never thought I would root for @sama but competition is very much needed, and luckily they delivered. GPT 5.6 totally delivered so far, and it seems like it's gonna be our new best friend for a while. I have never seen a startup that is actively being hostile to their power users the way Anthropic is. Actually blows my mind. Well.. so long ✌️

AI will replace workers almost everywhere. In gaming it goes the other way. Wrote about this in @khaleejtimes this week. The general rule: where the value of work lives in the result - not the human - AI takes the job. You don’t open a spreadsheet and admire the accountant's taste. Most of the economy is built like this, which is why most of the economy is about to feel it. Games however are not built like this. You cannot keep the result and lose the person. The result IS the person. Taste, tension, why one idea is fun and its neighbour is dead on arrival…that judgment IS the product. Everything underneath it is a production line. The barrier was never judgment, it was grind. @AngryBirds was Rovio's 52nd game and still took a small team eight months and $100K. That wall keeps millions of creators out. Most of the field is using AI to tear the wall down by deleting the creator too. Prompt in, "game" out, hollow lifeless games the end product. We are building @SPARQWORLDS on the opposite bet: the machine takes the production line, the creator keeps the game. Full piece below.

what is your target of daily launches on the Neural Launchpad ? we're not aiming for hundreds of token launches a day. We believe we're better off finding volume in higher-quality tokens than just a mass push of token launches we believe that's a healthier model for the ecosystem

SPARQ is almost here🚀

🧵> NEURAL x SPARQ update livestream we did a quick stream yesterday to give an update on where things stand with both SPARQ and Neural covered SPARQ’s next phase, the upcoming closed beta, our content plans, and how we’re approaching the web3 side moving forward quick thread on the main points 👇

See you on Steam🎮