Summary
The founder of Yorby — an AI-powered social media marketing tool for startups — breaks down every line item of operating a $120K ARR SaaS business, arriving at a total monthly infrastructure cost of roughly $4,198. The video is a detailed, real-money look at what it actually costs to run an AI product in production, covering Supabase ($600/month for database and storage), Vercel ($150/month for hosting, with one painful $1,000 spike from serving video assets incorrectly), Anthropic’s Claude API ($220/month for text generation), Google Gemini plus Google Cloud Platform services for video and image processing, and several other tools.
Yorby’s two main features — an AI content script writer that remixes existing viral social media videos and an AI content studio that generates fully synthetic marketing content — lean heavily on large language models. The founder notes that Claude usage dropped significantly after pivoting away from individual creators toward B2B customers, while Gemini handles the heavier media processing workloads. The video also discusses using AI coding assistants (Cursor and Codex exclusively — no Claude Code) and how AI has replaced what would previously have required a dedicated DevOps hire for Google Cloud Run deployments.
The second half addresses potential cost optimizations: migrating off Vercel to self-hosted infrastructure, consolidating Supabase usage, and whether the current cost structure is sustainable as the product scales. For developers and founders evaluating AI API spend at the indie or early-stage level, the video offers rare transparency into real production numbers.
📺 Source: Your Average Tech Bro · Published July 29, 2026
🏷️ Format: Workflow Case Study
