Foundry

About Foundry

Inspiration

The UGC (User-Generated Content) advertising market is exploding, but creating high-quality ad content remains expensive and time-consuming. Small businesses and solo entrepreneurs often can't afford $500-$5000 per video from professional creators. We asked ourselves: What if AI could democratize UGC creation?

Foundry was born from this vision: to give anyone the power to generate scroll-stopping TikTok ads in minutes, not weeks.

What We Built

Foundry is an end-to-end AI-powered platform that transforms a simple product description into polished TikTok ad creatives:

Smart Script Generation - AI analyzes your product and generates persuasive, hook-driven scripts optimized for social media
Custom Visual Creation - Generates product images and lifestyle shots tailored to your brand
Automated Video Production - Combines scripts and visuals into ready-to-post video ads
Iterate & Refine - Real-time feedback and refinement tools to perfect your creative

All powered by a seamless payment system using Echo Protocol, allowing users to pay-as-they-go without subscriptions.

How We Built It

Tech Stack:

Frontend: Next.js 15, React, TypeScript, Tailwind CSS
AI Integration: Google Vertex AI for script generation, Imagen for visual creation, video synthesis
Payments: Echo Protocol for crypto-native microtransactions
State Management: Custom React hooks for complex generation workflows
UI/UX: Radix UI components for accessible, polished interfaces

Architecture Highlights:

Asynchronous video generation with real-time status polling
Modular component system separating image, script, and video generation
Client-side history management with localStorage
RESTful API routes handling AI model orchestration

Challenges We Faced

Async Video Generation - Video synthesis takes 30-60 seconds. We built a robust polling system with status checks and error handling to provide smooth UX during long-running operations.
AI Prompt Engineering - Crafting prompts that consistently produce high-quality, on-brand scripts and images required extensive iteration and testing.
Cost Management - Integrating crypto payments via Echo Protocol while ensuring fair pricing and preventing abuse was complex but rewarding.
State Synchronization - Managing the flow between script generation → image creation → video synthesis while keeping UI responsive required careful orchestration with custom React hooks.
Error Recovery - Building graceful fallbacks when AI models fail or timeout, ensuring users never lose their progress.

What We Learned

AI Model Orchestration: How to chain multiple AI services (LLM, image gen, video gen) into cohesive workflows
Crypto Payments: Integrating Web3 payment rails into modern web apps
Async UX Patterns: Creating delightful experiences around long-running AI operations
Production-Ready AI Apps: The gap between "AI demo" and "reliable product" is huge (error handling, rate limits, and cost management are critical)