About
We build AI tools that return hours to video professionals.
ContentIndexer is an AI video analysis API used by editors, wedding photographers, eLearning creators, and developers building automation on top of video.
Last updated:
Why we exist
Video editing is still dominated by manual scrubbing. Editors spend hours searching long-form footage for the dialogue, shot changes, and key moments that will make the final cut. Wedding photographers cull twelve-hour days a frame at a time. Training creators stitch together chapters by hand.
We think that's solvable. ContentIndexer runs state-of-the-art vision and audio models against your video and returns structured metadata — the kind a computer can reason about and your editing software can import. Faces, speaker turns, shot changes, CLIP embeddings, and vertical-specific moments like first-dance detection or code-screen tracking. You get OTIO, FCPXML, or CSV exports that drop straight into Premiere, Final Cut, or DaVinci Resolve.
What we do
- AI video analysis API. Submit a YouTube URL or uploaded video. Get faces, objects, pose, shot changes, transcription with speaker diarization, and CLIP embeddings back as structured JSON.
- Vertical presets. Wedding, video editor, ZoomCraft (coding tutorials), and training — each preset bundles detectors and domain-specific moment detection.
- NLE-ready exports. OpenTimelineIO, Final Cut Pro XML, and CSV — import directly into every major editor.
- Integrations. REST API, webhooks (HMAC-signed), Zapier, SCORM export for LMS platforms, and a Model Context Protocol server so Claude and other AI assistants can use the platform natively.
How the platform works
Under the hood, ContentIndexer is serverless. Videos are chunked and fanned out to parallel AWS Lambda workers running MediaPipe, CLIP ViT-B/16, FFmpeg, and AssemblyAI. Results aggregate into S3, and a lightweight Cloudflare Worker + D1 database handle auth, billing, and the public API. The system is designed to scale horizontally — a twelve-hour wedding processes in the same wall-clock time as a thirty-second clip.
Meet the founder
Scott Havird
Founder & Engineer
Scott builds ContentIndexer end-to-end — from the Python Lambda inference pipeline to the React marketing site you're reading now. His background is in full-stack engineering and video tooling for creators and training teams.
Contact
Questions, feature requests, partnership ideas? Reach us at [email protected].