Differo: AI Website Change Monitoring

The Challenge

Building a production SaaS normally takes a whole team across backend, frontend, infrastructure, and billing. The goal with Differo was to prove that a lean, senior team, with AI-accelerated development, could ship a complete, production-grade product end to end without cutting architectural corners.

Our Solution

Differo was architected and built from scratch: a .NET 9 backend following Clean Architecture and CQRS, a React and TypeScript frontend, and Cosmos DB for storage, all running on Azure App Service. Azure Functions handle scheduled monitoring, headless-browser scraping with Playwright, and webhook delivery. An Azure OpenAI analysis layer explains what changed on a page and whether it matters. Stripe is integrated end to end: checkout, customer portal, and a webhook handler for four event types. AI wrote roughly 90% of the code while every architecture decision and every line was reviewed.

Impact

• Complete production SaaS designed, built, and operated by OldSport
• AI wrote roughly 90% of the code, with every line reviewed
• Serverless architecture runs at about $0.03/day in production
• Stripe billing live end to end: checkout, customer portal, and webhooks

Tech Stack