AI Tools at Vidio

How Vidio (Indonesia's #1 Streaming Platform) Built Great Engineering Culture—Now Supercharged by AI

Topic 9 of 17

Published Apr 6, 2026 · Clip 36:17–44:20 (8:03)

Tommy shares a failure — a text-to-SQL chatbot that failed because users lost trust after errors and preferred accountable data analysts — and successes: AI models ('anak intern') detecting ad breaks in video streams, sentiment detection, and metadata labeling. In development, he traces a progression from unimpressive GitHub Copilot, to a step-change with Cline's memory-bank feedback loop, to Claude Code, and using AI to apply patterns consistently across services. The DevOps team built MCPs for infrastructure documentation and the data lake, and he emphasizes grounding AI work in test-driven development.