Case Studies
A glimpse of what happens when production constraints disappear.
Mix Reality with AI
From a $20k, 2–3 week VFX-heavy animation process to a 2 hour AI-native build. Anchored in real training footage.
Exercises morphing into machine equivalents. Generated with AI-powered workflows directing FFLF, camera movements, and morphs.
Problem
Traditional character-morph fitness spots need a studio, athlete, lighting crew, VFX artist, animator, compositor, editor and colorist. They cost at least $20k and tie up the team for 2–3 weeks, mostly spent in post-production and animation.
Strategy
Film exercise reps at the gym to get maching first- and last-frame references, then direct AI video models to generate morphs and camera movements. No influencer? No problem. Frontier video models support Motion Control, so you can entierly replace the person in the video with a character image reference.
Solution
AI-native workflow: 30 minutes of filming to get first- and last-frame references, then focused workflows and prompting using ComfyUI and Kling. The result is a scroll-stopping sequence with real camera paths, mind-bending equipment transformations, and a turnaround measured in hours instead of weeks.
AI-powered gym in Santa Monica that uses adaptive equipment and personalized data to optimize fitness training.
AI vs. VFX
From a 2 week, $5k–$10k editorial shoot with specialized VFX tools like Blender to an AI-native build with prompt engineering.
Room-scale iridescent bubbles suspended above a sandy beach. An impossible environment created in minutes.
Problem
A single fashion spot requires $5k–$10k and at least two weeks. You are coordinating booking the model, stylist, makeup, photographer, location, permits, transportation, and lighting crew. Then comes post-production: VFX, editor, and revisions.
Strategy
What used to be gatekept by specialized VFX tools is now limited only by your imagination. Treat concepts as brand worlds instead of one-off shoots. Use AI to design impossible environments that feel on-brand for your product and audience. Play around with different concepts that weren't accessible before.
Solution
AI-native workflows: creative brief, product images, styling references, prompt engineering, and a few rounds of iteration. The result is an editorial campaign that would have required a full production team.
Impossible Photoshoots
From a $500k, 6–8 week multi-location campaign to a 3 hour AI-native build across impossible locations.
You want to send the models, where? Epic and impossible landscapes. Lava flows, cliffs, glaciers, all in one show-stopping campaign.
Problem
A multi-location fashion campaign means flights, accommodation, permits, transportation, crew, styling, VFX and post. Budgets land between $150k and $280k and production drags for 6–8 weeks. May the odds be with you on perfect weather.
Strategy
Design a brand world that can hold impossible environments you'd never be able to shoot. Lock the product, theme, and color story, then let AI handle the landscapes, lighting, and camera while you focus on narrative and styling.
Solution
AI-native workflow: mood boards, wardrobe references and environment prompts. In roughly 3 hours you have a six look, multi-location campaign that would have required an international shoot schedule drowned in paperwork and logistics.
Technical Work
Applications, AI Agents, and Projects
Content Lab
Agent
Solving video production bottlenecks with AI. Agentic orchestration of Blender, Remotion, and ComfyUI to create complex, high-quality videos in minutes.
AI Tools: Gemini, Antigravity, Cursor, Claude, Codex
Stack: Next.js, Python, Blender, Remotion, ComfyUI
OpenClaw
Agent
Clawdia is your 24/7 personal AI assistant. She communicates via Telegram, with dedicated topic threads for contacts, calendar, expenses, and news. Contact capture flow: I take a picture of a business card and she adds the info to my contacts, researches person and company, and drafts a follow up email. Built on OpenClaw, running on a private VPS, locked down with Tailscale, and designed with hard limits on what she can do without asking first.
AI Tools: GPT-5.4, Claude, Gemini, OpenRouter
Stack: Python, Telegram Bot API, Google APIs, Hetzner VPS, Tailscale
Calibrate
Agent
Confidently automate your workflows. Upload your workflow and let AI analyze its automation compatibility. Create your agent org chart, A2A cards, and tool registries. This was a team submission for the Google x Kaggle Agents Intensive Program in Nov 2025.
AI Tools: Codex, Claude, Cursor, Google ADK, Gemini, NotebookLM
Stack: Next.js (React), Python services (FastAPI), Firebase Auth + Firestore
Curately
App
Your taste profile, built around you. Skip the $300 appointment and upload two photos to get your face, body shape, and color analysis in seconds. Send your AI agent shopper to find products and assess compatibility for you: add your holy grail products (and your never-agains) across beauty, skincare, hair, and cosmetics. The more you tell it, the smarter your recommendations get. Fashion module coming soon.
AI Tools: Gemini, Claude, Codex, Cursor
Stack: Next.js, React, TypeScript, Supabase, Vercel
Plant Daddy
App
Never let another plant die on your watch. Powered by Gemini 3 Livestream video and voice. Show him your plants and he'll figure out environment, health status, and create custom care plans live on the call. Created for a Deepmind Hackathon, coming soon as a mobile app.
AI Tools: Gemini, Antigravity, Cursor, Claude, Codex
Stack: Next.js, React, TypeScript, Firebase Auth + Firestore
FitnessAI
App
Set your goals and log workouts/PRs. If you use another logging app, AI parses your screenshots. Get custom progress analysis and weekly workout plans based on recent history. In-depth charts for progressive overload, imbalance detection, and calorie burn. It's like having AI Tableau for your workout data.
AI Tools: Gemini, Genkit, Firebase Studio, Antigravity, Cursor, Codex, Claude Code
Stack: Next.js (React), TypeScript, Firebase Auth + Firestore
Macros Tracker
App
Chat with your AI coach and have it calculate your macros, log your meals, and give you live advice on how to hit your macro goals. View daily, weekly, and monthly rollups to stay on track.
AI Tools: Google Generative AI, Cursor, Antigravity, Codex, Gemini
Stack: Next.js (React), TypeScript, Firebase Auth + Firestore, Vercel
reimagen Website
Other
Created with AI IDEs, CLI coding agents, and GenAI video models. Video assets were created with Sora, Meta AI, Veo, and via OpenArt's platform to battle-test the models from the big international labs: Kling, WAN, Seedance, Minimax Hailuo, PixVerse, Vidu, etc. TLDR: Do not sleep on these, especially Kling.
AI Tools: Cursor, Codex, Antigravity, Sora, OpenArt
Stack: React (Next.js), TypeScript, Vercel
Custom GPTs
[ Requires ChatGPT Account ]
YC Office Hours
GPT
YC partner board simulation giving you direct, high-leverage startup feedback. Challenges assumptions, asks clarifying questions, and adapts feedback by company stage.
Start Chat →a16z Office Hours
GPT
A16Z-style advisory board simulation partner-level strategic feedback. Provides differentiated takes across product, market, and defensibility.
Start Chat →AI Creative Director
GPT
A creative partner that turns your ideas and aesthetic intent into structured, model-optimized prompts designed to improve composition, motion, and visual consistency.
Start Chat →Social Copywriter
GPT
Writes platform-optimized copy for TikTok, IG, Facebook, YouTube, LinkedIn, and X. Adapts tone and structure per platform while enforcing basic compliance and brand safety.
Start Chat →Focus Group
GPT
Runs simulated user reactions across key demographics and social platforms. Surfaces conflicting viewpoints instead of averaging responses.
Start Chat →FitnessAI Companion Coach
GPT
AI fitness companion paired with FitnessAI for training and recovery insights. Interprets uploaded InBody data conservatively and avoids medical claims.
Start Chat →


