Building a new feature for my app | coding vlog ep. 3

Quick Overview

The founder decided to pivot the initial plan of launching a hiring tool first and instead focused on building the Viral Content Database feature for their app, Yorby AI, to better serve the social media marketing space, even though the initial development phase was longer and more technically complex than anticipated due to needing a robust backend and handling numerous edge cases.

Key Points: The founder decided to pivot from building a hiring tool to prioritizing the Viral Content Database feature for Yorby AI, aiming to solve social media marketing problems. The development of the Viral Content Database feature took approximately one month of deep work, which was longer than the initial two-week estimate. The core technology enabling this feature is leveraging the Gemini 1.5 Pro API for advanced content analysis and remixing. The system architecture for content scraping involves an Admin User initiating a workflow via a Next.js Admin App, which triggers an Upstash Workflow to handle verification and scraping logic against various platforms (TikTok, Instagram). The engineer is currently testing the system with 1-2 videos but plans to test with 100-200 pieces of content next week to ensure scalability and check for errors. The founder strongly advises other founders to talk to users early, citing that external feedback (even if sometimes contradictory) is crucial for product direction, which is why they are launching a minimal feature first.

Context: This video is the third episode in a startup vlog series documenting the development of an app called Yorby AI, which aims to be a comprehensive social media marketing tool. The founder is providing an update on the progress made over the past week, specifically focusing on overcoming challenges related to a new content inspiration database feature and the need to incorporate user feedback early in the development cycle.

Detailed Analysis

The founder kicks off the vlog by stating the current date (Thursday, October 16th) and confirming the immediate goal: building out the first major component of Yorby AI, which is the Viral Content Database. This tool is designed to help social media marketers find high-performing viral content for inspiration, allowing them to query content, check if it exists, and then use an AI (Gemini 1.5 Pro) to remix that content for their own needs. The system design, shown via a sequence diagram, involves the Admin User sending video URLs to the Next.js Admin App, which triggers an Upstash Workflow. This workflow verifies admin privileges, then loops through each URL, scraping data (platform, content ID) and checking if the content already exists in the Supabase DB before processing. The entire process is built on a serverless architecture using Vercel and Next.js. The founder admits the initial development took about a month, longer than the two-week estimate, due to complexity, especially around handling edge cases and ensuring resilience against timeouts when scraping large batches of content. A key lesson learned is the importance of getting user feedback early, contrasting with past experiences where too much external input caused distraction. The founder has successfully tested the core scraping logic with one or two videos and plans to test with 100-200 videos next week. The video concludes with a sponsor mention for Warp, an AI-powered terminal, and a brief personal reflection on balancing hard work with taking time off for friends and family.