I Tested Manus AI: China's Wrapper Beats OpenAI (Plus, Vibe Coding For Creatives)
China's wrapper agent outperforms OpenAI Operator, creatives get coding superpowers, and the commercial tools reshaping the 3D & VFX landscape.
Welcome To The March Edition 🌍✨
The frontier is expanding faster than ever. This month we've got agents redefining autonomy, creative superpowers you didn't know you had, and a future where every pixel is generated, not rendered. Let's dive in.
Manus AI: China's Autonomous Agent 🤖
I tested Manus AI. It's the closest thing I've experienced to a truly autonomous AI agent. They’ve basically wrapped Claude, Qwen and a few other lego bricks to deliver a product that blows the $200/month ChatGPT operator out of the water.
Hopefully this release pushes big US labs to release less hobbled versions of their own autonomous AI agents. I can't wait till this thing can use desktop apps like Premiere and Photoshop to really let us automate x-tool workflows.
Watch my 10 minute review & analysis on YouTube (and X)
Now, I wouldn’t call Manus a DeepSeek moment (like many are) — instead it’s more of a Baby AGI moment. Yet, the range of reactions are fascinating:
“It’s just a wrapper” to their co-founder basically saying - “yea, so?”
Creative Superpowers You're Not Using Yet ✨
No code? No problem. LLMs like Claude, Gemini, Grok and ChatGPT have gotten a huge boost lately, yet most folks just use them to write and summarize text. Let’s get into vibe coding for creatives!
I walk through building a 3D city simulation, dynamic video annotations, smart shot lists, and even a futuristic AR HUD overlay using nothing but prompts. This is about rethinking your relationship with AI as a creative partner, not just a tool.
Watch LLMs for Creatives in 42 minutes on YouTube
YouTube Chapter Links:
(00:00) LLM Super Powers You're Not Using
(00:26) Claude 3.7 Overview (No Coding Required)
(02:38) 3D City Vibe Coding for Creatives
(10:01) Shot List Editor & Visualizations w/ Claude
(13:26) Lines Between Code & Content Are Blurring
(15:08) Google AI Studio for Multimodal Workflows
(19:34) Annotating Videos w/ Gemini & After Effects
(23:03) Automatically Select Best Takes w/ AI Studio
(25:38) Script & Delivery Feedback w/ AI Studio
(29:22) Deep Research Is Wildly Powerful
(32:08) How I Use AI To Learn New Things!
(35:12) Grok 3 for Better Thumbnails & Outlines
(37:54) How To Use Multiple AI Models Together
(40:32) Vibe Coding for Creators
The Tools Changing 3D & VFX Today 🎬
Here’s how Claude, Gemini, Grok, and ChatGPT can effortlessly 10x your creative workflow—today.
I’ll walk you through how I built a 3D city simulation, dynamic video annotations, smart shot lists, and even a futuristic HUD overlay using just AI prompts. Plus, how to use AI models as your creative collaborator.
If you're a creator looking to save hours, produce better work, and simplify complexity, this video is for you. The focus is practical — tools you can use today with clear commercial-use licensing.
YouTube Chapter Links:
(00:00) Introduction
(00:34) AI has entered the chat...
(01:37) Making VFX: The Old Way vs. The AI-Powered Way
(02:43) Spatial Intelligence: AI 3D Tools You Need to Know
(18:53) Visual Intelligence: The AI Revolution in VFX
(28:43) Generative AI & The Future of Creative Production
(38:47) Conclusion & What's Next
Creative Tech Quick Hits ⚡
My Forbes Interview on AI video: Had a blast joining my creator friends to discuss 2 big AI video models — Veo 2 vs Sora — on the heels of its UK release.
Reality Reconstruction: Shawshank Redemption prison reconstructed in breathtaking 3D using real-time techniques ([link])
Vibe Coding in Blender: The lines between code and content continue to blur, creating new possibilities for 3D artists
Runway's Restyle Video Launch: This new reskinning tool vastly outperforms competitors like Viggle — it even did a great job with me playing guitar.
Asteria/Paul Trillo Masterpiece: The stunning blend of classical and generative AI techniques points to where visual creation is headed.
Generated, Not Rendered: Jensen was right. We're rapidly approaching a world where every pixel is generated rather than rendered.
GPT-4.5's Quiet Landing: Despite the hype, the reception has been surprisingly muted. What does this tell us about the trajectory of LLMs?
Apple Intelligence Delayed Again: The rollout continues to face challenges, raising questions about Apple's AI strategy. Here’s my 2c.
VRML Dreams Reborn: Today's vibe coding renaissance feels like the spatial web dreams of the 90s finally finding their true form.
Driving A Car With A Keyboard: This engineer clearly loves PC gaming so much he just wants to WSAD through life.
On The Horizon 🔭
YouTubers are descending upon South by Southwest, and I'm excited to share that I've been curating a killer set of talks and demos for TED2025. Look out for more details in the next edition!
If you found this useful, share it with fellow reality mappers. The future's too interesting to navigate alone.
Cheers,
Bilawal Sidhu
https://bilawal.ai