AI ยท Media

MetadatAI

Turn video into structured, usable data

MetadatAI: a video goes in, AI extracts transcripts, people, scenes, objects, on-screen text and a summary, a private node network does the processing, and structured data comes out through an API.

Project Info

Platform

Distributed AI media processing, API-first

Runs On

Dedicated GPU servers and Apple Silicon

Technologies

Vue 3, Node.js, Fastify, MongoDB, Python, MLX, GPU inference, local AI models, OCR, speech recognition, REST APIs, webhooks

Engineering Highlights

  • Distributed node architecture
  • Scales by adding nodes
  • Local AI on GPU or Apple Silicon
  • Configurable processing pipelines
  • API-first, with webhooks
  • Multi-tenant organizations and permissions
  • Live job and pipeline visibility

Video In. Structured Data Out.

MetadatAI is a distributed AI platform that breaks video into its parts, then serves what it finds through an API, so other apps and workflows can use what used to be locked inside the file.

  • Transcripts & timestamps
  • Speakers
  • Scenes
  • People & faces
  • Objects
  • On-screen text (OCR)
  • Summaries
  • Custom prompt outputs

The Challenge

Most software treats video as a single file. Knowing what was said, what was shown and who appeared usually takes several systems and heavy infrastructure. And for sensitive or proprietary media, sending it to outside AI services raises privacy and cost concerns.

A Private Network of AI Nodes

Instead of one central server, MetadatAI links private AI nodes, on GPU servers or Apple Silicon, into one coordinated platform. Sensitive media can stay on hardware the organization controls.

Need more capacity? Add a node. A deployment can start with a few machines and grow with demand, without redesigning anything. More nodes. More capacity. Same platform.

Pipelines You Configure

Reusable templates decide how each video is analyzed: transcription, speaker analysis, vision, face recognition, OCR, custom prompts and summaries. An app submits the media and a template, and MetadatAI runs the rest.

Built as Infrastructure

MetadatAI is a processing layer, not a closed app. Other systems submit media, track each job and pick up the results, which makes it a fit behind asset management systems, production workflows, content platforms, media archives, internal tools and AI agents.

Have a Project in Mind?

Let's discuss how we can help bring your vision to life with our expertise in software development.