🎉 Depictio 1.0: from prototype to production
Depictio went live last year, and it worked. People built real dashboards with it and shared them. It was also slow: past a few tens of thousands of rows it went from sluggish to unusable, and there were enough rough edges that using it meant working around them.
The idea never changed: point Depictio at the outputs of a pipeline, tell it how those datasets relate, and get a dashboard where filtering one thing filters everything. What changed is that it now does that at the scale your pipelines actually produce, and at speed.
That is what 1.0 means here. Not "feature complete", which no tool ever is, but four specific things:
- It stopped moving. The data model, the API and the viewer are stable, so a project you ingest today still opens next year and code you write against the API does not need rewriting at the next release.
- It is meant to be deployed. Independently scaling API and worker pools, Docker or Kubernetes, and health reporting when an ingestion goes wrong, instead of a thing you babysit.
- It holds up on real data. Millions of rows across linked collections, on hardware you already have.
- A lot is already in it. Eighteen advanced visualisation types, MultiQC rendered inline, images, maps and trees, not a roadmap promising them.
Here is what changed.
🎬 Depictio end to end: a project, its data collections, and the dashboard built on top.