Medium potentialLow license · MITScore 10/10

VideoLens

Let LLMs actually watch your videos

642PythonBuild: 4-6 daysHUANGCHIHHUNGLeo/claude-real-video

VideoLens turns any video into a concise, scene‑aware set of frames and a transcript that LLMs can read. It runs locally, so you control what data leaves your machine, and it’s ready to plug into Claude, ChatGPT, Gemini, or any other LLM.

Tech stack

Next.js 14 App Router + Docker + Vercel + Stripe + Clerk

BYOK model

Not required

Freemium model

Free tier: 5 videos/month, 30 minutes total, no audio extraction. Pro tier: unlimited videos, audio extraction, contact sheets, priority support.

Deployment complexity

Medium — Requires ffmpeg, yt-dlp, and optional GPU for Whisper, but can be containerized and deployed on a single server

SEO keywords
LLM video summarizationvideo keyframe extractionscene-aware video analysisLLM video processing toolvideo summarization with Claude
Why it qualifies
  • High demand from content creators for quick video insights
  • Unique scene‑aware extraction reduces token costs for LLMs
  • MIT license allows commercial use with minimal friction
  • Low hosting cost – can run on a single GPU server
Build this
ShareXLinkedInHN

Related analyses