Early development · 1.0.0 not released yet

Your media,
understood.

MetaLoom is an open source media asset platform. Point it at a folder and it transcribes, detects, describes and indexes what is inside — on hardware you control.

See what it does Read the docs

Project status

Not released yet.

MetaLoom is in active development. The scope of 1.0.0 is written down and the demo container runs today, but nothing is tagged as a release, and interfaces can still move.

What 1.0.0 will contain Follow the build log Join the Discord
Version in tree
1.0.0-SNAPSHOT
Published artifacts
none yet
Demo container
runs today

Two ways in

Where would you like to start?

No jargon

I have an archive I cannot search.

A short, visual tour of what MetaLoom finds in your footage and how you get it back out.

Take the tour APIs and pipelines

I want to run and integrate it.

Architecture, node catalogue, REST and GraphQL reference, deployment playbooks, Helm charts.

Open the docs

What it is

One platform, four jobs.

01

Take in

Reference files where they already live, or upload them. Nothing is copied to a cloud you do not own.

  • filesystem & upload
  • content-addressed
  • libraries
 02

Understand

Speech becomes text, faces become vectors, scenes become captions, documents become searchable content.

  • 19 node kinds
  • GPU-friendly
  • custom nodes
 03

Find & ask

Full-text and vector search over everything that came out — or just ask the built-in agent.

  • Elasticsearch · Qdrant
  • chat agent
  • skills & memory
 04

Run anywhere

A single container on a workstation, or a scaled worker fleet on Kubernetes. Same platform.

  • Docker · Helm
  • REST · GraphQL
  • web · desktop · CLI

All features

Built on
  • Java 25
  • Vert.x 5
  • PostgreSQL
  • jOOQ
  • React
  • GraalVM native
  • Kubernetes
  • whisper.cpp
  • OpenCV
  • Apache 2.0

From the build log

How it is being built.

All posts