Summary
Creator Riley Brown walks through building a fully functional team web app using ChatGPT Work and the ChatGPT Sites feature, claiming the resulting tool replaced three SaaS subscriptions costing thousands of dollars per month — built in three prompts with no coding experience required. The app, called Social Scrape, pulls short-form video content from Instagram and TikTok, stores stats and transcripts, and hosts them in an authenticated, team-accessible web application.
The tutorial covers the full prompt-to-deployment workflow: crafting a detailed initial prompt specifying the app’s layout, user authentication, database storage, video playback, and integration with the Scrape Creator API skill; running ChatGPT Work for 25+ minutes on the desktop app; and iterating via in-browser annotations to fix UI issues such as video sizing, profile images, and download button styling. Brown demonstrates how the desktop app’s in-browser annotation system lets users leave comments directly on the running app, which Work then uses to make targeted changes to the live deployment.
The video explains ChatGPT Sites’ built-in infrastructure — hosting, database, authentication, file storage, and custom domain support — and positions it as the midpoint between standard ChatGPT and Codex, OpenAI’s developer-focused agent platform. The broader pitch is aimed at non-technical creators and small teams looking to build custom, shareable internal tools without hiring developers or paying for SaaS platforms, using natural language prompts as the primary interface.
📺 Source: Riley Brown · Published August 26, 2026
🏷️ Format: Tutorial Demo
