// what devs are shipping
Feed
Stuff people are actually building today — bugs, wins, half-finished side projects.
1
All of this came from member feedback, so keep it coming.
Build Log: DevConnectAll of this came from member feedback, so keep it coming.
Freelance is live.
Freelance and contract are their own type on Jobs now, with a filter.
One click and you only see project work instead of scrolling past full time roles.
Listings carry fixed price versus hourly, duration, scope, tech stack and who the client is.
Budget stays mandatory. No budget, no listing. Clients can post project briefs directly too.
Projects is live.
A new page to browse what people here are shipping, tech stack on every card, filter by the stack you care about. No more digging through the feed to find the good stuff.
The site got a lot faster.
Images were the problem.
The logo alone was shipping at 746 KB on every page, and photos were served at full resolution.
Everything is properly sized now, images load with the first paint instead of after, and pages preload as you hover a link.
Should feel noticeably quicker, especially on mobile.
Post your project with its stack so it shows up in the filters, and tell me what still feels slow or off.
1
How Icelanders are thinking about AI
A look at how Icelanders are approaching AI, featuring insights from a national AI education pilot for teachers. The video explores public perspectives and the implications of integrating AI tools in education. It reflects on Iceland’s early government-led efforts to shape AI literacy and policy.
5
I Built This Shooter With Opus 5 AI (Full Match Gameplay)
Everything shown here was developed over 4 days using Claude Opus 5 together with Blender, MCP, and OpenAI image generation.
I only used Opus on high the whole time.
This isn't an AI showcase.
It's a real game being built with AI as a development tool.
FAQ
Q: What engine is this?
Godot 4.
Q: Did AI build everything?
Claude Opus 5 wrote all of the code.
I created the specifications, tested every feature, and directed the project.
Q: Are the bullets hitscan?
No. Every bullet is physically simulated with travel time, penetration, ricochets, and energy loss.
Q: Is this multiplayer?
Yes.
Q: How long did it take?
4 days.
4
Opus 5: What I Learned in Three Days, Not a Perfect Demo
Three days with Opus 5, never above "high" reasoning, and a 9mm I fired ricocheted off a steel door and killed me.
Nobody scripted that.
This is Project Pool, a shooter built in Godot 4.
Every "I built a game with AI" video I watch stops at a cube bouncing on a grey plane, so this one goes the other way: what the systems actually do, how I asked for them, and where the model fell apart.
WHAT'S IN IT
Ballistics. Every round is a real object with mass and travel time. No hitscan, so you lead your shots.
Impact angle decides penetration, deflection or ricochet. Caliber decides how much energy there is to spend, and whatever survives the wall is what reaches the person behind it.
Explosives. Grenades throw individually simulated fragments instead of applying a damage radius.
Half a concrete wall saves you, plywood doesn't.
Every fragment path is drawable in a debug view.
Bodies. Ragdolls driven by impact point and momentum. Where you hit someone changes how they fall.
The world. A full interior level with real lighting, sunlight through windows and shadows that move.
Every room specced with exact wall coordinates and ceiling heights before a single mesh existed.
The pipeline. Assets modelled in Blender, driven by the model through MCP. Textures and concept art through the OpenAI API.
HOW I ASKED
1. Specs before code, with real numbers. Wall coordinates, ceiling heights, material resistance values. Not "make it realistic".
2. Read the existing code before writing new code.
Most AI-project spaghetti is a model rebuilding what already exists.
3. A definition of done, plus a test that enforces it. My level geometry only stopped leaking once a script was firing rays out of the map hunting for holes.
4. Every taste decision stayed mine.
The ballistics were correct on the first pass and felt wrong. Correct and good are different targets.
4
We want your feedback!
We want your feedback!
DevConnect is built for developers, so your opinion shapes where we go next.
What do you like so far? What's missing or needs improvement?
Drop your thoughts in the comments. Every reply gets read
5
What's new: a big batch of updates
Build Log: DevConnectHey everyone.
A big batch of updates just shipped, and most of it came straight from your feedback, so thank you.
What's new and where to find it:
AI tools (menu, top right → AI tools): Startup Radar analyzes your idea, Repo Health reviews any public GitHub repo, and Time Machine replays a repo's journey.
AI twin: open any profile and hit "Interview AI twin" to ask about their work.
Find your people (menu → Discover): Find collaborators and the new For You page match you with co-founders and complementary builders.
Video posts: bigger uploads, thumbnails, a real player with seek and volume, plus cinema mode and fullscreen.
Chat: reply to any message, pin your important chats, and the AI helper now sits right at the top of your chat list.
Company pages: real reviews and salary reports, opened from any job's company name.
Profiles: case studies, project metrics, live-demo buttons and a reputation score based on what you actually build.
It's all in the menu, top right. Free, and no ads.
Seriously, thank you for the ideas and the bug reports.
They shape this directly. Keep them coming, reply here or ping the AI helper in chat.
4
I upgraded my Osu Mania 3D Clone
Build Log: TETHRANow the multiplayer is working.
I also improved the graphics a lot :)
It's a 3D rhythm game.
I also added a auto update version via Git so users don't need to reinstall the game.
3
I Turned osu!mania Into a 3D Rhythm Game
Build Log: TETHRAThetra is a fast-paced 3D rhythm game built for players who care about precision, timing, and flow.
Inspired by the gameplay that made keyboard rhythm games so addictive, Thetra reimagines the experience with a completely new 3D visual style and its own identity.
This video showcases the current state of development, gameplay, and the direction of the project.
If you're into rhythm games, feedback is always welcome.
follow the project for updates.
4
Rebuilding osu!mania in pygame with a weak AI
Before Opus 4.8, I spent two months rebuilding osu!mania, the rhythm game, from scratch in pygame. No game engine, no framework, all of it built with AI that was a lot weaker than what we have now.
Quick context for anyone who hasn't played it: osu!mania is a keyboard rhythm game. Notes fall down a set of vertical columns, and you press the matching key at the exact moment each note reaches the hit line, in time with the music. Think Guitar Hero or DDR, but on your keyboard. Simple to describe, brutal to get right.
I didn't build it to ship a product. I built it to actually learn: how far you can push a weak model, where the APIs and context limits fall apart, and how much of "good design" still has to come from you when the AI can't carry it.
The hardest parts, in order:
1. Timing. In a rhythm game everything has to land exactly on the beat, and "almost" is worse than useless. Keeping the audio, the input and the falling notes in sync down to the millisecond is where most of the two months went.
2. Accuracy scoring. Every note you hit gets judged by how close you were to perfect timing (perfect, good, ok, miss) using real timing windows, and that feeds a live accuracy percentage. Getting those windows and the math to feel fair, exactly like the real game, took way more iteration than I expected.
3. The UI and buttons. pygame hands you a blank window and nothing else. No buttons, no menus, no UI toolkit at all. Every clickable element, every menu, every hover and press state I had to build by hand. Back then, with weaker AI, even "make this button reliably do the thing" was a real fight.
The part I'm proudest of: it imports songs straight from real .osz files, so you can drop in actual osu! beatmaps and play them, charts and audio and all.
I keep seeing "I made a game with AI" posts where the game is a colored square that moves. Nothing wrong with starting there, we all start somewhere. But I wanted to see what happens when you point even a weak model at something genuinely hard and refuse to stop at the first thing that runs. Turns out, a lot. The tool being weak just meant I had to understand every part myself, which was kind of the whole point.
3
I built a 3D game engine with AI before Opus 4.8 and learned a lot
A while back, before Opus 4.8 dropped, I gave myself a challenge: build a real 3D game engine from scratch with AI, and see how far I could actually push it.
It was never meant to ship. No product, no users, no roadmap. Just me trying to learn what building with AI really feels like when you go deep on something genuinely hard.
Three weeks later I had a working 3D engine. Plenty of it broke along the way, and I rewrote things more times than I'd like to admit. But that's exactly where the learning was. AI didn't hand me an engine. It handed me a thousand decisions I still had to make and understand myself.
Posting a video of it here because this is what DevConnect is for: real things people build with AI, rough edges and all.
Curious from anyone who's tried something similar: what's the hardest thing you've built with AI just to learn, not to ship?
1
Resuelve background update
Any ideas to improve my background
And my login section
With High quality vector pictures / graphics
Like Google or Apple
Thanks for your help
🤝
2
The different levels of how Claude thinks
Explains how Claude processes information in layers, drawing parallels to the global workspace theory. The video introduces concepts like J-space, silent math, automatic processing, and hidden thoughts, illustrating different levels of conscious-like thinking in an AI. It highlights how certain representations become publicly accessible within Claude’s internal workspace.
3
I'm looking for people who would like to help me with my project.
I'm looking for testers to help me find bugs and errors.
I am also looking for employees for permanent positions.
I'm looking for someone who can develop games
and would like to help me with my project. If you have any questions, feel free to send me a message.
My Project: Resuelvedr.com
Thank you for your time
2
What tools do u use to code ?
what MCPs/Tools/AI do u use to code mor efficent ?
3
Working on website update
Today I work on a Website Update for this Website.
Get it more seo optimized + optimize the mobile version.
What do u think is better :
Dark or white mode ?
2
RESUELVE ALL IN ONE APP
I am currently developing an all in one app for the Dominican Republic and have found excellent partners in the Vyren team / DevConnect to support me with security.
0
My (not so pretty) journey in tech
TL;DR: A developer shares a candid, non-linear journey through C++, data structures and algorithms, machine learning, and web development, emphasizing perseverance and realistic expectations in learning tech skills.
The author describes feeling overwhelmed learning C++, then pushing through despite setbacks. They critique the hype around success stories and highlight that failure is a normal part of growth. They note DSA requires sustained effort and that learning timelines vary by person. They share that ML sparked curiosity across multiple tools and libraries, and that the desire to explore led to broad interest in tech, including web development.
Question for the room: What realistic learning milestones or timelines have you found effective when tackling multiple tech domains at once?
— via dev.to
0
The Most Dangerous AI-Generated Code Is the Code That Passes All Tests
TL;DR: A developer reflects on AI-generated code that passes all tests but introduces hidden issues due to untested relationships and architectural assumptions. The core lesson is that passing tests don’t guarantee correctness of design, invariants, or future integrity.
The author notes that AI-written code can seem correct because tests pass, yet still break unseen relationships and invariants. Tests verify only those behaviors you chose to check, not broader system health. Everything appeared green: CI, tests, lint, build, and the AI's explanation, but the real risk lay in gaps between what was tested and what the system requires. The story warns that reliance on passing tests can lull teams into false confidence about code quality.
Question for the room: Have you encountered a case where tests passed but a later change revealed a deeper architectural flaw?
— via dev.to
0
I Tried Building a Website With AI From My Phone
What happened when I decided a laptop wasn't a requirement anymore It started on a Tuesday...
— via dev.to
