The Problem
Modern SaaS tools for productivity and automation can quickly become expensive, especially for solo developers or small teams trying to orchestrate complex AI workflows. The challenge was to build a robust, scalable system to automate content generation and data processing without relying on paid tiers of services like Zapier or AWS.
The Architecture
I architected a serverless, decoupled system using three primary components:
1. Google Colab (Compute): Acted as our heavy-lifting engine for running AI models and complex Python scripts for free. We exposed it via ngrok for webhook triggers. 2. Supabase (Database/Auth): Provided a generous free tier for PostgreSQL and real-time subscriptions, acting as the central nervous system connecting the frontend to the backend. 3. Chrome Extensions (Client): Custom lightweight extensions served as the UI to trigger workflows directly from the browser, injecting data into Supabase.
1. Google Colab (Compute): Acted as our heavy-lifting engine for running AI models and complex Python scripts for free. We exposed it via ngrok for webhook triggers. 2. Supabase (Database/Auth): Provided a generous free tier for PostgreSQL and real-time subscriptions, acting as the central nervous system connecting the frontend to the backend. 3. Chrome Extensions (Client): Custom lightweight extensions served as the UI to trigger workflows directly from the browser, injecting data into Supabase.
The Execution
By listening to Postgres changes in Supabase, the Colab notebooks could immediately pick up new tasks queued from the Chrome Extension. Once a video or text generation task finished, Colab updated the row, and Supabase's real-time channels instantly notified the frontend.
The Impact
This entirely free stack successfully processed over 5,000 automated tasks in its first month, scaling digital content production by 90% and saving an estimated $300/month in SaaS subscriptions.