Latest YouTube Videos

The AI Coding Workflow I Used to Build Nebula

https://github.com/agentSystemLabs/nebula

A real review on GPT 6 Astra (Not Another 3D Demo)

After using it for 24 hours, I've never been so excited to build and explore new things. Thank you Firecrawl for sponsoring the video, check them out: https://rasmic.link/firecrawl Bezalel: https://bezalel.sh/ Eve: https://github.com/michaelshimeles/eve-agents Adam: https://github.com/michaelshimeles/adam Connect with me: My Site: https://rasmic.link/site Follow me on X: http://rasmic.link/x Follow me on Instagram: https://rasmic.link/instagram Follow me on Tiktok: https://rasmic.link/tiktok Join my Discord: https://rasmic.link/discord Subscribe to my second channel: https://rasmic.link/more-micky

Is GPT-6 Astra Worth Cancelling Your Other Subs

GPT-6 Astra just came out, almost nobody has hands on with it yet, and everyone is trying to figure out if it's worth cancelling their other subscriptions for. So I'm going to find out the only way that actually means anything, which is building real apps with it live, in depth, until we hit a wall. OpenAI shipped Astra on September 3 and the benchmarks look incredible on paper, and I want to know what happens after hour two on a real project when the context gets messy and the thing has to hold a whole codebase in its head. We're starting with [GAME CONCEPT] and going from there, and I'll tell you straight where it's better than what I'm already paying for and where it isn't. Come hang out in the chat, throw ideas at it, and let's see where it falls over. I spent 12 years as an engineer at Apple, and now I build apps live on stream with AI tools and share the whole process as I go. Join the free community where we're all building together: https://rfer.me/startmyai (My small live cohort opens a few times a year, and you'll hear about it first inside the community.) Timestamps 00:00:00 GPT-6 Astra Introduction 00:01:25 Game Concept: Mario Galaxy & Katamari 00:04:48 AI Thumbnail Generation Process 00:07:08 Game Mechanics & World Building Ideas 00:10:21 Massive Initial Prompt to Astra 00:12:55 Astra's Clarifying Questions for Game Design 00:18:50 Astra's Initial Tech Stack Recommendation 00:22:27 Eric Provenche's Guide to Prompting Astra 00:27:44 Astra's First Detailed Game Plan 00:29:49 Shifting to an Apple-Native Tech Stack 00:36:23 Visualizing the Core Gameplay Loop 00:44:10 Verifying Blender Access with Computer Use 00:48:15 Astra Controlling Blender Live 00:54:42 Starting the Full Game Build on Ultra 01:00:05 Astra's Multi-Agent Building Process 01:05:50 Stream Crash and Recovery 01:11:35 Showcasing the 'Cere' Transcription App 01:17:50 First Live Playable Demo of the Game 01:28:30 Polished Concept Art from ImageGen 01:39:20 OpenAI's Full Usage Reset Announcement 01:40:40 Canceling Anthropic Subscription Live 01:46:14 Giving Feedback for a Polished Version 01:52:00 Banked vs. Regular Usage Resets 01:56:46 Astra's Polished Game Proposal 02:02:10 Reviewing New 3D Assets in Blender 02:08:00 Final Prompt for Polished Game 02:18:20 Code Architecture of the Cere App 02:28:30 Full Starter Prompt Recap 02:37:00 Second Playable Demo: New World & Gameplay 02:49:00 Astra's Low Token Usage Analysis 02:55:40 Using Grok Bot to Automate Community Posts 03:01:20 Final Thoughts and Stream Wrap-Up

This Model Ruined Every Other Model For Me

GPT-6 Astra is my favorite model of all time and it's not even close. This thing has absurd capabilities in computer use, swarming, and more that make it truly second to none (even if it has some downsides) Thank you Firecrawl for sponsoring! Check them out at: https://davis7.link/firecrawl My X, Discord, and more: https://davis7.sh/

Getting a keyboard that THOCKS for FREE

because a thocky board is a basic human right?

Shadcn Just Made Tailwind 30x Faster

shadcn and Aiden just released `cn` — a new engine for Tailwind class merging and conflict resolution that replaces `clsx` + `tailwind-merge`. Same API. Drop-in replacement. Zero dependencies. And up to 30× faster for common component calls. In this video, I break down what `cn` is, how it works, why it’s so much faster, and how you can migrate an existing shadcn/Tailwind project. 🔗 Check out `cn` on GitHub: https://github.com/shadcn-ui/cn This video is sponsored by ShadcnBlocks. Check them out 👉 https://www.shadcnblocks.com/ 🎁 20% discount code: ORCDEV I build real products in public and show you exactly how I do it. Want the stuff that doesn't make it into the videos? Free newsletter: https://orcdev.com ━━━━━━━━━━━━━━━━━━━━━━━━━ 📬 THE NEWSLETTER — free The exact tricks, workflows, and tools I use to ship real products. What worked, what broke, and how I fixed it. No fluff. → https://orcdev.com 🚢 SHIPPER CLUB For people who actually ship instead of just planning to. → https://shipper.club/ 🎥 VIDEORC This video was made with it. → https://videorc.com/ ━━━━━━━━━━━━━━━━━━━━━━━━━ FIND ME GitHub → https://github.com/TheOrcDev X → https://x.com/orcdev #orcdev #shadcn #ai

From AI Slop to Pro Design: MuseSpark 1.3 & Gemini 3.8 Flash

Most AI-generated websites end up looking like generic, low-effort slop. In this video, we test whether a specialized coding skill—Impeccable—alongside automated browser testing (Ego Browser) can transform raw, clunky web pages into polished, modern frontends. https://www.bitdoze.com/best-ai-skills-for-web-design/ We put Gemini 3.8 Flash and Muse Spark 1.3 head-to-head to see which LLM follows instructions, executes UI refactoring best, and delivers a layout you’d actually ship.

Gemini 3.8 Flash vs Muse Spark 1.3: Don't Trust the Benchmarks!

AI benchmarks don't tell the full story. In this video, I put the newly released Gemini 3.8 Flash and Meta’s Muse Spark 1.3 to a real-world coding test: building a complete multi-page Astro website from scratch with custom SVGs and clean architecture. Ai Design skills: https://www.bitdoze.com/best-ai-skills-for-web-design/ We compare their code organization, speed, visual output, and how they handle autonomous browser verification—without using custom design skills. Watch until the end to see the final websites side-by-side and which model actually took the win.

Herdr: The AI Agent Emulator You Can't Ignore

Discover how to orchestrate autonomous AI coding agents, manage remote VPS sessions over SSH, and ensure your terminal tasks keep running in the background—even after you close the window or disconnect. Article: https://www.bitdoze.com/herdr-agent-multiplexer/ GLM 10% oFF: https://z.ai/subscribe?ic=NKNUNYDRZT Timestamps: 0:00 - Introduction to Herdr vs. Tmux 1:40 - Terminal UI, Mouse Support & Adding Spaces 2:19 - Automated Agent Orchestration with Herdr Skills 5:42 - Background Sessions: Closing & Resuming Without Data Loss 7:02 - Connecting to a Remote VPS for 24/7 Autonomous Tasks 9:15 - Herdr Mac App Overview & Zed Editor Integration 12:10 - Final Thoughts & Workflow Recap

I have a software factory, it’s just all skills (you can copy them)

In this video, I share with you how I ship features at speed while maintaining quality using my software factory Thank you Cursor for sponsoring the video, check them out: https://rasmic.link/cursor My skills: https://github.com/michaelshimeles/skills Bezalel: https://bezalel.sh Connect with me: My Site: https://rasmic.link/site Follow me on X: http://rasmic.link/x Follow me on Instagram: https://rasmic.link/instagram Follow me on Tiktok: https://rasmic.link/tiktok Join my Discord: https://rasmic.link/discord Subscribe to my second channel: https://rasmic.link/more-micky

I Threw Away My App and Lost Almost Nothing

I threw away the whole app and rebuilt it from scratch, and the only reason I didn't lose much is that I wrote the spec down first. I had the Mac app working and connected to the cloud, and then I realized I don't want to charge people a subscription for something the new local models can just run on their own device, so I had Cursor read everything and generate a spec file, and that spec is the recipe I'm using to make the software again. I handed it to Bitrig, which drives my Claude Code and Codex subscriptions and keeps everything on the SwiftUI rails, and it builds in phases so each phase gets verified before it moves on, and right now the live mic and the demo are working. I'm in Hawaii for Hawaii Tech Week this week so there's more coming every day, and the whole point of this app is one flat price, no subscription, buy it once and keep it forever. Not sponsored by Bitrig, I pay for it and I just really like it. I spent 12 years as an engineer at Apple, and now I build apps live on stream with AI tools and share the whole process as I go. Join the free community where we're all building together: https://rfer.me/startmyai (The AI Builder Club, my small live cohort, opens a few times a year, and you'll hear about it first inside the community.) Chapters 00:00 What Happened After the Rewrite 00:38 I Don't Want to Charge Subscriptions 01:15 What Bitrig Actually Is 02:07 Why I Break It Into Phases 03:03 Keeping the Code on the Rails 04:20 The Live Mic and the Demo Are Working 05:05 Buy Once, Keep Forever

The ULTIMATE Hacking Keyboard