Prompt Flow

💡 Inspiration

As a solo developer working extensively with AI, I kept running into the same issue: people didn’t know what to prompt or how to reuse good prompts. Most interfaces give you infinite possibilities, but no direction. I wanted to create a system that felt structured, dependable, and capable of turning messy prompting into composable workflows.

PromptFlow was born from that gap. It’s designed to help anyone, not just power users, chain together perfect prompts and use them in production workflows, while enabling creators to earn from their contributions.

⚙️ What it does

PromptFlow is a platform where users can:

• Discover curated, ready-to-run AI prompts
  
• Chain prompts into multi-step flows
  
• Track credit usage per inference
  
• Monetize prompts and flows (set fees, earn payouts)
  
• Connect outputs to automation tools like n8n (coming soon)

All usage is tracked via credits, with a shared ledger for purchases, debits, and creator payouts.

🏗️ How I built it

• Frontend: Next.js + TypeScript + Tailwind CSS
  
• Authentication: Supabase Auth (Google + GitHub OAuth)
  
• Database: Supabase (PostgreSQL)
  
• Backend Integration: Sonar API for AI model execution
  
• Payments: Stripe for credit purchases
  
• Credits: 1 unified ledger table for all credit transactions
  
• Deployment: Vercel
  

Everything is event-driven. There’s no polling, no mock data, no fallbacks, only real, production-grade execution.

🧱 Challenges I ran into

• Designing a robust credit system that supports usage, bonuses, refunds, and creator payouts
  
• Keeping the database schema lean but flexible for flows, prompts, and payments
  
• Coordinating prompt execution and credit deduction in a way that’s fair, traceable, and developer-friendly
  
• Managing Supabase schemas in parallel with auth providers like Google

🏆 Accomplishments I'm proud of

• Building an actual production-ready app, not just a demo
  
• Avoiding fallback code, everything works for real users, with real authentication and real transactions
  
• Creating a reusable architecture that will scale with additional AI models and automation tools

📘 What I learned

Finalizing requirements before involving LLMs is crucial. AI is powerful, but only as good as the instructions it’s given. The more detailed the plan, the more productive AI becomes in helping you build.

🚀 What's next for PromptFlow

• Prompt Flows: let users chain together multiple prompts into composable flows
  
• External Automation Hooks: n8n-style automations to turn outputs into actions
  
• Credit Subscriptions & Bonus Programs: reward power users and creators
  
• Prompt Templates: editable, cloneable blueprints for faster building
  
• Public Profiles: showcase top creators and earners
  
• More AI Models: expand support beyond Sonar and OpenAI via Lilypad
  

PromptFlow isn’t just a prompt marketplace—it’s the foundation for a new way to think about AI as infrastructure.

Built With

• css
• next.js
• sonar-api
• stripe
• supabase
• tailwind
• typescript
• vercel