The user is looking for an API that turns videos into RAG-ready chunks, extracting transcript, speaker labels, sampling frames, running OCR, generating scene descriptions, and returning everything already embedded and chunked.
I'm working on a problem I kept seeing in my network: **RAG over video is annoying.** Most teams either: 1. Manually transcribe + chunk everything (nightmare) 2. Use heavy enterprise APIs (Azure Video Indexer, Google Cloud Vision) and deal with complex pricing 3. Stitch together Whisper + EasyOCR + embeddings themselves (reinventing the wheel) So I built **VectorVid**—an API that does the hard part: **What it does:** * Takes a video URL (YouTube, S3, etc.) * Extracts transcript + speaker labels * Samples frames and runs OCR (so you catch visual context: slides, UI, diagrams) * Generates scene descriptions * Returns everything **already embedded** and chunked What you get back: { "chunks": [ { "start_sec": 42, "end_sec": 68, "text": "Our pricing is $49/month...", "scene_description": "Slide showing pricing table", "ocr_text": "Starter $49 Pro $99 Enterprise Custom", "embedding": [0.12, 0.45, ...] } ] } Then YOU plug it into Pinecone, Weaviate, Supabase pgvector—whatever. You own the RAG pipeline, I handle the video understanding. **The demo:** I indexed a few sample videos (iPhone keynote, product demos, lectures). You can search inside them and see the exact output you'd get as an API response. **For RAG devs specifically:** * You don't have to tune your own chunking strategy for video. * OCR + transcript in one output (stops you from losing info in slides). * Timestamps are baked in (so you can link back to the source). **Early feedback I'm looking for:** * Is this the right level of abstraction? (Too much/too little?) * What would you want to customize? (Chunking strategy, OCR languages, etc.) * Pricing thoughts? (Considering \~$0.03-0.05/min indexed) Live demo + waitlist: [https://www.vector-vid.com/](https://www.vector-vid.com/) Would love RAG builder feedback. Comment or DM if you have a use case in mind.