Raw Animals: from idea to scheduled publishing
In Raw Animals I treated AI as part of a complete production line. One topic or file in Google Drive could start scripting, generation, assembly, copy and publishing, with a human checkpoint where it mattered.
ORCHESTRATION
n8n connects many small responsibilities
The workflow read from Drive or Sheets, passed context to AI agents, called external APIs, stored task identifiers and checked statuses. Delays, conditions and retries were necessary because video generation is not immediate.
PRODUCTION
From short clips to a finished format
I built variants for videos, AI image posts and Explained by Emoji with a custom Remotion renderer. A piece could pass through script, TTS narration, a safe 9:16 or 16:9 frame and final render before reaching Postiz.
RELIABILITY
The queue remembers what happened
For larger batches PostgreSQL stored topics, statuses, attempt counts, history and content fingerprints. This made it possible to resume after failure, route difficult cases to review and limit duplicates.
OPERATIONS
Containers are part of the solution
On a Linux server I operated n8n, Postiz, PostgreSQL, Redis, the renderer and helper services in Docker. I handled volumes, connections, schedules, file cleanup and safe API key storage.