DEV Community: UnitBuilds

Dwarven Stronghold Update!

UnitBuilds — Wed, 28 Oct 2026 07:22:48 +0000

For anyone excited, Dwarven Stronghold's sim loop is nearing completion and the procedural rendering engine is working!

Still alot of wiring to do (almost 10k modules for the sim-render) but it's getting there... It just took 10 million lines of Rust, but it's finally taking shape.

So whose excited?

If you are, lemme know, so once I actually get it at a playable state, you can test it for me (yes, be my guinea pig). Likely gunna be in a worse state than Minecraft's Alpha, but hey, that just means it can only improve!

Insane to think that a simple game takes longer than RustC to compile... But I guess that's to be expected for building it in Rust, compile slow, but it's tiny and mighty.

Sometimes the unexpected is incredible.

UnitBuilds — Sun, 25 Oct 2026 22:59:53 +0000

So lately I've been experimenting with a new system, I've been trying to reinvent the proverbial wheel of AI. LLMs are just too flaky and computationally expensive, I wanted to explore what I can do, to make it cheaper, not hallucinate and just work faster, offline.

So without going to deep into details, I started off with abstractions and rules, to try and build a translator, it kinda worked, it could take an english sentence and translate to german (tricky, cuz of how it assigns genders to words).

But I'm not a linguist, so I phased off to coding, something truly deterministic (that part's important). So I decided to go a bit outside of my norm and targeted python as my first trial. Essentially what I wanted to do, is feed it codebases and have it analyze and optimize it. So alot of tinkering later, it works... I fed it 5 codebases, it analyzes them in seconds, suggests parity functions from the other codebases that do the exact same thing, but better (a scale, still gunna make it preference adjustable, but needs way more data for that).

So what I'm currently at, isnt a LLM, it cant come up with a single thought of it's own... But what it can do, is optimize code faster than any LLM and with 100% certainty. No hallucinations (physically impossible), full auditability (it can citate the exact function that's more optimal), fully tested (verifies parity) and completely deterministic (you heard me, no matmul, no funky inference... Pure deterministic). And best part, no need to retrain it everytime you want to add to it, just add it.

Obviously all of those are pretty big claims and while the proof is there, it's at such a small scale, it cant even be trusted on it's own merit, I still need ALOT more data before I can claim anything with certainty. But it goes to show, you never know what you find down a rabbit hole... My next goal after I get python nailed down, is cross-language translation (imagine being able to convert any python codebase to rust in seconds?). Then feed it ALOT more data and then the final test... Chromium. Something about Chromium has always bugged me, I wanna crack it. 30m lines of code, hundreds of thousands of files, quite frankly, impossible for any LLM to handle, let alone effectively, or on a non google scale budget. Though I'm not getting my hopes up anytime soon... That behemoth wont be easy to even fit on my system, let alone compile. But what if.

What if it works... Chromium, the single thing guaranteed that every single person in the world makes use of, suddenly overnight 10% faster? It's a pipe dream, but we all need a challenge to get the best out of us. My first approach was the foundry (too expensive), then my IDE (too computationally heavy... and expensive). But this... No matmul, no inference, hell I dont even need a GPU for it! I dont even need a ton of ram for it either!!! It probably wont work, but hey, I'm excited to try and who knows, maybe this time it's just feasible enough to do.

So I've got some questions for all of you out there who've gone though open source codebases. Which is in your opinion, 'good'. I dont mean functionality wise, I mean when you read the code, it's like a savant wrote it, elegant, efficient, secure, durable, all the things we wish we wrote daily, but rarely achieve. Doesnt need to be massive, doesnt need to be python, it just needs to be good (atleast in your opinion). So I can feed it in and use it as training data. Currently, of the ones I analyzed, flask is the leader.

Update, things are coming together.

UnitBuilds — Thu, 22 Oct 2026 08:59:07 +0000

So it's been a while and there's been some developments...

Firstly, I joined YC startup school and connected with 4 potential co-founders (to work on their stuff). Then YC gave me an email that made me second guess myself. I read on YC, that companies with 2 founders are more likely to succeed. Then I got an email from them saying in summary: "stop finding someone to co-found with, if you just want them to market it. Learn marketing yourself, it's okay, we do fund solo founders".

So that, along with the webinars were eye opening... I built all of these amazing things, but never really focused on 1, I knew IDE was my main, but I never focused, why? Turns out, subconsciously I knew what I was doing when I made my company's website. At the time, the foundry was expensive to run and not very viable with models at the time, though I made it work. Times have changed and all I built was just parts of the greater scheme. I built my MCP server, to make a high-concurrency, dynamic MCP server that can run anything. I built Drone, so remote management becomes AI native. I built Windows Automata, to give true computer use control to AI. I made MCP-Lite to understand the AOM and what it could mean for AI browsers. I made my IDE, so I would never use it... Why need an IDE, if you build AI infrastructure, to let AI do all the work?

Turns out, that's the moat and the trajectory for my startup. Open source the developer tool, productize the true potential of it. Automated software generation, from prompt to binary, like my foundry was all these months ago. Except now, with much better models, much better tooling and way better scalability.

Cuz IDK about you, I wouldnt just outright trust blackbox AI generated software? But if I get the source code and can debug and inspect it myself... Suddenly it becomes a speedup vs manual POC creation.

Reading about other startup founders and their dilemma, they're all in the other boat, they all need 'someone technical to own engineering from 0-1', so why not give them a damn good AI team to prototype for them? You need a POC to approach investors, else it's a pipedream... Some already just prompt their way to a POC, but they do this over weeks/months, because AI as we use it, relies too much on feedback, despite the answer being obvious even to it, it's an artificial guardrail.

I've already had the core architecture sorted out months ago, what I was lacking, was a way to make it efficient and scalable. Now I have that. So next up, is applying for funding!

On another note, Dwarven Stronghold is officially over 10m LOC and is currently compiling (funny, it's hitting my poor system harder than compiling rustc) , so I can test it and make sure it all works! Visually, wont be impressive today, probably gunna be really buggy... But the simulation is what's important and it runs at 5x the tick speed of Dwarf Fortress, so I had to have done something right!

Other than that, Doccit, is about to get a hell of alot of work. We're aiming to have it take over operations entirely by end of the year for the entire firm!

Vitaling is done, but I reached out to some people in medical industry on YC, to co-found with me, so they can push the capabilities beyond a 'self-care' app and into a medical care app.

So it's been quite an interesting week... I'd highly encourage everyone, even if you're happy with a day-job, to go do YC's startup school, I thought I wouldnt learn much... I was wrong and chances are, you could learn a few things too, eg. how 'vesting equity' is a trap.

And if you're in the job-market, there's plenty of EXTREMELY wealthy startups that are looking for co-founders too, the 1 guy did EdTech, grossing 2.5m USD a MONTH! So he wants more people doing EdTech, so he can grow an enterprise. So why work a day-job, when you could be co-owner, which means you earn a salary + % profit of the business? And the day you leave, you keep your shares, so unlike a day-job, when you're done, you're not throwing away all your efforts, your efforts still generate an income for you.

I think we're cooked...

UnitBuilds — Sun, 18 Oct 2026 23:12:14 +0000

Honestly, tonight I had a realization moment. My token pan for qwen cloud expired, so I thought 'you know what, I wanna see what local AI can do'. For context, I'm working on my IDE, so I took my buildbox (ryzen 5500, 32gb DDR4, GTX 1070 8gb, Ubuntu Server), using Drone, let my AI set up an Ollama instance, running Qwen 3.6 35b A3B Q5. 22gb total, sits comfortably in ram, A3B sits comfortably in 8gb vram with room to spare... 35 TPS... Sounds like nothing, but considering a word average is probably around 1.2 tokens... Suddenly that becomes a sentence a second. I imagine practically everyone reading this, given we're developers and tend to put our money where it's earned, into hardware, likely has a comparable system? 35 TPS, is about 1/5th what you can expect from a cloud model, except this is running on the hardware you already own and it doesnt stop you from using cloud too... The reason it made me question life as I know it, is because 35b params used to be the models of legend, the things that only ran in cloud, because nobody could afford a GPU large enough to run it... Now 8gb of vram is enough to get quite reasonable performance out of it... It changed how I saw open weight models and the future of the entire industry. Go buy a 2nd hand RX 580 for $50, a scrap PC for $100, if ram wasnt so insanely expensive, $100 for 32gb of ram, now you added a new permanent node to your AI pool for $250 and it runs a damn good model, you can get away with half that if you go for a lighter model, or decide to offload execution to 1 system, instead of dispersing it to the nodes. Think about that, what you pay per year on AI subscriptions (or what some pay in 2 months), you can have a permanent AI added, no subscriptions and it doesnt stop at 1 node, add 10, add 100, it doesnt matter, the models are smart enough, that you'll barely miss cloud models.

I just read an article too on how Halo got ported with AI acceleration and improved, Adobe and Microsoft's suites are being redone in rust ALL AT ONCE! We're reaching the point where code isnt cheap anymore, it's worthlessly inexpensive, because for every name-brand, there's a vibe-coded knockoff that likely performs better... And that's scary as a developer, where are we all gunna be in 5 years? 'But enterprise companies pay for software', not if the knockoff is open source and they have AI too... To say we're screwed is an understatement. And maybe it's not a bad thing. The Halo port meant that the creative arts of making games is nearing it's final call, software companies are pushing open source, because trying to protect your code is impossible now, we're hitting the point where I honestly cant see what value any software can hold that a knockoff produced over a weekend doesnt. 'Infrastructure', it takes AI 10 minutes to get the same setup on Cloudflare, it's as secure, if not more than what you built. 'Quality', they set the agent up to patch it continuously. 'reliability', the rewrote it in rust, it takes less memory, runs faster and is memory safe by default. 'but mine looks better', they've used screenshots and replicated GUIs perfectly for years... What can we honestly provide today, that AI cant do better, cheaper, faster?

Even IT support is becoming a thing of the past, hell I built Drone exactly to do that, now I dont even troubleshoot myself anymore, I just let my AI do it for me, cuz it saves me the hassle. The part that should really scare you, is what comes after we're replaced. If it can rewrite adobe's suite, microsoft's suite, build an OS in 12 hours and write a complete ERP system in less than that, how much effort do you really think it takes to automate it's use? 'But surely entertainment is safe?', CGI replaced practical effects 20 years ago and AI generated content trends longer and harder than human created content today. Take Gangnam Style, it broke records... But that Ai generated trumpet sound used on TikTok... How many times do you think THAT's been heard? 1 billion times? Or closer to 10 billion times? The global economy has already shifted and we're all still playing catchup. The irony is, as anyone who works a desk job, where's our future?

Thank goodness it wont happen overnight and the water is barely coming to a boil, but if there's any advice I can give anyone, it's learn a physical skill for the day using a computer wont earn you a penny, because it might come sooner rather than later.

Sorry if all of this sounds super pessimistic, I imagine in a way, it's actually optimistic, if not for the fact that it means we're all on the chopping block... What are your thoughts?

We are no longer continuing the interview.

UnitBuilds — Mon, 05 Oct 2026 17:08:35 +0000

So late last night, I got an update from Wasmer, that they will no longer be continuing with my application. A bit sad, ngl, especially given friday wasnt by choice that I had to cancel the interview. But all the best to them.

So what now? I've been applying to more jobs and prepping my pitch deck, so I can look at funding UnitBuilds CC and V.E.L.O.C.I.T.Y. IDE! Which honestly would be amazing!

Right now, I'm taking a break though. Since I left my job, I've been non-stop busy and couldnt rest. So I decided I need a hiatus, rest and relax, then get back at it in a week or 2. In the meantime, I'm going to get back to my roots and relax, play some games (Tainted Grail has been REALLY fun!), went to the beach today, gunna do a BBQ now and go to bed, without the sound of 2 laptops and 2 computers running in the background.

Honestly, not very excited taking a break, but I think that's all the more reason to take 1. I wont stop developing during this time, infact, I think I'll be developing even more, but on the things that matter to me. IDE has gotten to a pretty solid place now, so it's minor refinements and testing. I DEFINITELY need to update my website (yeah, it's bad). Probably should release the health and fitness app, then beyond that, I'm not sure. Maybe refine pitch deck, set up cloud compute and model routing for the IDE, etc. That way it can live up to the dream I envisioned: "A lightweight IDE, that scales across all your available computers and if it has to, cloud". I kinda wanna refresh UnitBuilds itself (the original foundry) and open it up, I mean I have token plans to cover it, might as well open it up so people can test it.

Overall, I think I've gone too deep down the rabbit hole and I need some self-reflection time. Maybe finally read the Iliad and disconnect for a bit. But that's not quite me, I like software development way too much to quit cold turkey.

On other fronts, Doccit is doing well, I added regional support for practically most of the world, I wanna expand on it though. How many of you track your time and need time tracking apps like clockify to log your time. Imagine of the ide integrated with Doccit and automatically logged what you're busy with and where you spent that time... Optionally, with agent time tracking (not for billing, for production oversight), to see how much time the AI is doing the work and you're actually free to do whatever you like (lets face it, most of us either browse, or work on something else if we're invested).

So to finish up on this scatter brained spill of my mind. I have a question... If your IDE could do 1 thing, 1 magic thing, no matter how insane, what would it be? Genuinely, what is the thing you'd want most?

I had a few ideas, like a 'bottleneck detector' that highlights high traffic code that could be improved. Or a 'get it done' mode, where an AI team fully takes the product from 0 to 1, including comprehensive testing (not what passes for it by today's AI standards... Actual 24h+ usage to test it in real-world settings). Maybe a 'test suite' criteria checklist, where you state what you want tested and and what part of development, like E2E testing, integration testing and manual site traversal testing. I mean the IDE has full windows automation, so it could theoretically even use AutoCAD for you, if the model was smart enough to know what it's making. Or the core concept that UnitBuilds was founded on, 'prompt to binary', a true E2E production lifecycle in a single click, where your idea becomes a reality, no matter how many iterations it needs to go through to thoroughly complete and test it. Though as I experienced with the original foundry, sometimes, we're just a bit to vague on what we want and that leads to over or under engineering.

I'm sweating bullets

UnitBuilds — Thu, 01 Oct 2026 11:58:19 +0000

Well, about 4h30m to the interview and I'm stressed. Honestly, this is why I hate interviews, because it seems the more I study up, the more I feel like I dont know what I'm doing, even though I built it...

Mostly, because during the tech interview, I was asked really in-depth questions and expected to talk through implementations, quoting method signatures, which isnt really something I consider a skill of mine...

Not to mention, last time was just with the Tech Lead... This time it's with the CEO, Tech Lead and a Senior Dev... So a simple 'I did it this way, because that's the right way to do it' wont cut it. I need to be able to explain why I chose it vs what alternatives could have been used. All things I dont generally give a hoot for.

So needless to say, I'm staring down a barrel of a gun and I know I'm gunna dance like a chicken and make a fool of myself. Idk about any of you, but I work on multiple systems at once and I dont much pay attention to anything in particular, more the overarching architecture and making sure things actually work, are tested and benchmarked. If the benchmarks are unsatisfactory, then I optimize, simple as that. I dont much consider alternatives until I have the time to explore them and I dont consider them till they're tested and benchmarked. Not quite the kinda thing that shines in an interview...

What makes it worse, is it's an hour... For the AFS and the pgrust port combined... They wont beat around the bush, they're gunna hammer me with hard questions that I know I'll choke on. That being said, I have an ace up my sleeve... My SASOS can finally run and render native Notepad++. If I can find a gap to show them that, it might distract from the fact that I'm scatter brained and cant explain why --net during build is the exact fix needed 2 layers down, in order to make pgrust print times accurately. I'm stressed and I'm anxious and rightfully so, because I feel like I dont know what I'm doing. It's kinda like a carpenter who learned by trade, being asked 'what's this chisel' and they just call it 'thin chisel'. I can write solid code, but dont ask me to explain it, just say thank you that it works and it's tested.

In other news, IDE got a major upgrade. I added nodes! You can now host a headless running on an ubuntu server, then from your GUI instance, allocate work to be done on that system, so if you have a few old systems lying around and you wanna boost your productivity, you can now work on 1, execute on many. Next is cloud VM scaling, where a cloud instance can trigger expansions (think Kubernetes) for each agent it spawns, so ephemeral state agents can execute in their own sandbox environment, while maintaining connection to the sitemap. That way like you have a local LLM and a cloud LLM, you can just 'offload to cloud' for heavy builds or if you really want to power through a complex system via the teams studio.

Idea for that 1 being, wasmer's edge is actually perfect for it... Think about it, if an agent spawns as a .wasm instance, you can scale thousands of agents and scale back down to 0, cuz each agent is ephemeral, they just have their skillset and objective.

See... Scatter brained, I'm meant to be stressing about the debrief and I'm talking about my side projects! It just feels to me like I'm suited for working, not for interviewing... Cuz these are the kinda things that add value on-the-job, not before you land the job. They're looking for competence and I'm giving them experimental 'I'm not quite sure's' and that's what stresses me. Relaxing didnt help much either, clearing my mind came up foggy at best and I worry that I'm gunna trip up on the finish line. It took them a day to review the work... In that day, as the people who maintain the system, they most definitely knew exactly where to dig to find the flaws and the things I never thought of. An 'I dont know' or 'well that's why you're the lead' wont work... This time the interview isnt just with the lead (he's nice, during manual code review, he actually gave me hints, cuz he wanted to help), this time, it's him + 2... I dont know the other 2... I dont know if they're entering constructive, or destructive, if their goal is to weed me out, or to see my potential. Cuz I'm pretty sure I'm not the only candidate in the running at this phase. Probably am the only 1 who clocked in less hours than their predicted minimum though... And only 1 who logged AI hours at over 100h... But those can count to my benefit, as much as they can count against me. What I see as optimization, they maybe see as lazy ai prompting if I cant explain it.

So now that I've had my existential crisis and I've accepted my fate, come what may, I've got 4h left to prepare for my impending doom... Social anxiety's a bitch aint it.

Wish me luck (I'm cooked...)

Tomorrow is review day

UnitBuilds — Wed, 30 Sep 2026 13:06:03 +0000

So update for anyone following the journey. I finished the hiring task submissions and sent them through yesterday at 3am. Today, I got the email for the final review scheduled meeting. 6pm CET tomorrow. This is what the entire hiring pipeline boiled down to, this final review.

I'd be lying if I said I wasnt stressed, or anxious, but I think I got this. After all, I did build the systems they're gunna be asking questions about, not like it's a blackbox file out of context again.

So fingers crossed and lets see how it goes! I'll write an update once the review is done on how it went and when it's all said and done, I'll write an overarching post, detailing the hiring process, so people who dont know what to expect when going into the hiring pipeline have a reference point and know what are the things that I tripped up on and what are the things that really impressed them, so you know where to spend your time and what skills to refresh on.

I finished at 3am...

UnitBuilds — Tue, 29 Sep 2026 10:46:35 +0000

So I've been doing the Wasmer hiring tasks. As expected, they grossly underestimated the time required for the pgrust port... So why did it take me 8 days, but less than 4 hours?

To answer that question, you need to look at how I work. I built my own IDE, specifically because agentic orchestration made too many mistakes. So I fixed those mistakes, by building the infrastructure myself.

Now on to the fun part, it took me 2 hours of me-time to actually spec out the gameplan, find all the dependencies, chart it out into a roadmap and set up the team to do the task. Then it all boiled down to compile times.

I dont know if they'll hire me, but 7 PRs, 2 of which were production bug fixes. A fully working pgrust port published and tests and benchmarks to prove it. I dont know what more I could have done.

I finished the assignments and did the final submission at 3am today (10 hours ago), then went to bed. Woke up to my email being shared in the Slack channel and not a word since.

I honestly dont know what more I could have done. Yet I still feel it wasnt enough... And I think that comes down to the lack of ownership you feel with AI generated code. I signed off on everything, every edit was by my design, yet I just spent 4 hours reviewing, correcting, guiding, planning, never doing. So I guess it comes down to what do they value.

I think this is called progress?

UnitBuilds — Mon, 28 Sep 2026 06:33:21 +0000

So I've been working on Dwarven Stronghold for a while now. At one point, an overnight agent wrote a few thousand files as templates... This is the templates being fleshed out, file by file.

I think fair to say, it's getting to be a bit bigger than Dwarf Fortress, fingers crossed it's as good!

Also if anyone's curious, current LOC for the entire project is at:
Total Files: 9,652
Total lines: 10,519,594

I think they're angry?

UnitBuilds — Sun, 27 Sep 2026 15:50:05 +0000

Lmao, I think I won? 82 identical comments... 82 comments from me warning people. Comment 83: Nice bro.

https://dev.to/dev_supports - Please report ASAP, phishing scam and the anti-bot isnt detecting it!

UnitBuilds — Sun, 27 Sep 2026 15:10:25 +0000

DEV SUPPORTS - DEV Community

dev.to

Please confirm this is real

UnitBuilds — Sun, 27 Sep 2026 14:42:37 +0000

DEV.to team, please confirm this account is actually yours. As it's sending links for verification 'due to increased bot activity' in the comment sections. Kinda sus because it's a brand new account...