PRISM — AI Video-Content Platform
Short-form video pipeline: script (LLM) → narration (TTS) → assembly from stock → per-platform formatting → publish → per-video analytics
Founder, architect & solo developer
solo · solo
- Python
- LLM
- Cartesia TTS
- HeyGen
- Pexels API
- Pixabay API
PRISM — project creation (YouTube / TikTok / Reels)
Expand the deep-dive ↓
What doesn't work
Producing short-form video across multiple platforms and languages by hand is slow and doesn't scale: scripting, narration, footage selection, cutting, formatting and publishing eat hours per clip.
Architectural approach
A single PRISM pipeline: idea → script (LLM) → TTS narration → scene assembly from licensed footage → captions and formatting for YouTube/TikTok/Reels → per-channel publishing → analytics feedback. A human steps in only at the review stage.
Founder, architect & solo developer
Designed and built the whole platform: LLM script generation, TTS integrations (Cartesia, HeyGen) and stock footage (Pexels, Pixabay, Coverr), scene assembly, a review stage, a per-channel publishing scheduler, and per-video analytics collection (views, retention, engagement).