Will Your Reel Go Viral? Test It Before You Post
Posting a reel blind is a coin flip — you don't know if the hook lands until it's already out there. A quick walk-through of how to test whether a reel will actually land before you hit publish, so the duds stay in drafts and only the ones that earn the click ship.
I Built an AI Image Composer Because Editing Through Generators Is a Pain
Editing photos through AI generators is a pain — endless re-rolls, lost faces, broken lighting, wrong angles. So I built my own image editor on top of Qwen Image Edit 2511 that actually does what you tell it: edit a single image, rotate it to any camera angle, plug in custom LoRAs, or compose three separate photos (a person, a product, a background) into one seamless, photorealistic shot. Walk-through of every mode end to end — Edit vs Compose, Fast vs Quality, seeds, negative prompts, the multi-angle LoRA (SKS trigger word), a 3D camera-angle helper, custom .safetensor LoRAs, and a full compose demo. Runs locally or free on Hugging Face, fully open source. Heads-up: it's a 20B-parameter model that uses 2× GPU generation on HF Spaces and eats your quota twice as fast — watch the warning section before you run it.
Claude Soma — Running ~10 AI Workers on One Free Cloud Server With Just a Claude Subscription, All From Telegram
Claude Code is the brain. Telegram is the body. Claude Soma adds voice, memory, dashboards, tools, background workers, and autonomous AI teams that can keep working while you sleep. I show how I run around 10 AI workers on one free cloud server using just a Claude subscription — handling invoices, presentations, research, content creation, debugging, dashboards, and more, all controlled from my phone. Project: https://soma.mayankgupta.in
I Built a PRO AI Music Studio for FREE (Better than Suno?)
Try Ace Music Studio right now in your browser or run it locally! Today, we are moving from AI images and video into SOUND. I have built an open-source, full-stack AI song generator powered by the ACE-Step 1.5 XL model. Whether you want to generate full tracks with vocals, separate stems for karaoke, or apply custom LoRAs to completely change a song's genre, Ace Music Studio puts the power of a DAW right into a simple UI. No credits. No subscriptions. Just pure imagination.
Fix Ugly AI Images for FREE — Z-Image Studio
Paid image tools gate the good stuff behind signups and watermarks — so I built Z-Image Studio, a free browser-based playground running the new Z-Image model (Base + Turbo) with everything in one place: Text-to-Image, custom LoRA styles (Studio-Ghibli + fantasy), ControlNet (Canny, Depth, Pose), and an AI upscaler that actually adds detail. Walk-through of every mode — dreamy Ghibli and epic fantasy LoRAs, seeds explained, before/after upscaling, and Canny/Depth/Pose on both generated art and real photos — plus how to train your own LoRA to put yourself in any scene. 100% private: your images and LoRAs never leave your session. Free on Hugging Face.
I Built a FREE Kling AI Killer (And You Can Use It Right Now)
Tools like Kling, Higgsfield, and Veo are paywalled — so I built LTX AIO Generator, a free open-source playground on top of LTX 2.3 that does text-to-video, image-to-video, audio-to-video, lip sync, keyframe interpolation, and style transfer in one place, with native audio generation baked in. Walk-through of every mode (including a digital-you lip sync demo and an Iron Man + Wonder Woman style transfer), plus deep-dives on presets, seeds, negative prompts, LoRA adapters, camera movements, and Union vs Motion Track conditioning. Try it free on Hugging Face or clone the repo to self-host.
I Made Trump Say Anything With AI — 100% Free Voice Cloning in 22 Minutes
Voice cloning is locked behind paywalls — so I built Chatterbox Voice Studio, a free browser-based tool with three models, paralinguistic tags (laugh, sigh, gasp), multilingual support, and a dialogue mode. Walk-through of cleaning Trump audio with ElevenLabs Voice Isolator, cloning his voice in Chatterbox, side-by-side comparisons with the real audio, adding emotion, multilingual Hindi cloning, and an AI dialogue interview — all running locally with nothing stored on the server.
Chatterbox Voice Studio — Free AI Voice Cloning in Your Browser
Browser-based voice cloning studio built on the open-source Chatterbox model. Three models, paralinguistic tags (laugh, sigh, gasp), multilingual support, and a dialogue mode for full conversations with a cloned voice. Stores nothing on the server — runs free on Hugging Face Spaces.
Prompt Engineer — Custom GPT
A custom GPT that turns rough ideas into precise, structured prompts — covering role, context, constraints, and output format. Built to get higher-quality results from any LLM with less back-and-forth.
[WINDOWS] Set Up Claude AI in 30 Minutes — Research, Browse, Learn Anything
The Windows companion to the Mac setup video — install Claude Code via PowerShell, fix the PATH, wire up Playwright and Sequential Thinking MCPs, and drive it remotely from your phone. Includes live demos researching Nifty fluctuations, side-by-side good vs bad prompting with a custom Prompt Engineer GPT, and pasting screenshots into the TUI for visual feedback.
Ditched Linktree — Built a Smarter Links Page with Claude Code
Episode 2: Default YouTube share links open a browser on mobile, not the app. I build my own smart links page on my portfolio in about 30 minutes using Claude Code — with full Google Analytics tracking, a free portfolio cross-promo bonus, and auto-deploy via Vercel. Walk-through of MCP-driven planning, picking a client-side redirect for analytics, iterating UI with Claude, and raising the PR.
Set Up Claude AI in 30 Minutes — Research, Browse, Learn Anything
Episode 1: Install Claude AI (Desktop + Code), configure MCPs like Playwright & Sequential Thinking, and turn Claude into a personal research assistant you can control from your phone. Live demos include web browsing for Nifty stock analysis, chain-of-thought reasoning, and building a personalized YouTube learning plan from scratch.