DopCast started from a simple question: could a team of agents produce a motorsport podcast end to end? One agent researches the race weekend, one plans the episode, one writes the script, one voices it, and one masters the audio. It has a CLI, a Streamlit web interface and an API. Scripts also export to Markdown and PDF.
It taught me what multi-agent really means. The interesting work isn’t in any single agent. It’s in the handoffs: what each agent passes to the next, and what happens when one produces something the next can’t use. The voice went through three text-to-speech engines, pyttsx3, then gTTS, then ElevenLabs, before it sounded like a person.
In April 2025 I rewrote the pipeline around LangGraph, as an explicit graph of steps instead of a coordinating agent, and it got simpler. The rewrite also added Gemini for scripts, YouTube transcript extraction for research, and FFmpeg for mastering. That was the last work on it. The README still describes the older coordinator design. I wrote about it in A year in code.