Agnostic Data Labs

Dev diary

See your classifications at a glance with chips

Roelant Vos 2 min read

Classifications in ADL are core to the management of your design metadata. How you tag a piece of metadata drives to a large extent what it actually represents, and through which perspectives you want to visualise and interact with the metadata.

A column might hold Personally Identifiable Information (PII), a table might belong to the staging layer, a connection might point at sensitive data. Useful information — but only if you can see it where you are working.

That is what classification chips are for.

Classification chips rendered on nodes in the ADL directed graph

When chips are enabled, the classifications appears directly on the nodes in the model view and the directed graph. No drilling into a properties panel, no cross-referencing a separate screen. The label is right there on the node.

The interesting part is how you decide which classifications show up.

Rather than rendering every chip on every node — which would quickly turn the graph into noise — you configure this through personas. A persona is a collection of configurations tailored to a role or task: a compliance reviewer, a data architect, a developer working on a specific layer.

Each persona picks which classifications it cares about, and the chips render accordingly.

Persona configuration screen in ADL, selecting which classifications appear as chips

A few examples of how this plays out in practice:

  • A compliance persona highlights PII, Sensitive, and Confidential chips so reviewers can spot data that needs extra controls at a glance.
  • An architecture persona surfaces layer classifications like Source, Staging, or Integration to make the shape of the solution visible in the graph.
  • A delivery persona might hide classifications altogether, keeping the canvas clean while modeling.

Because personas are configuration, not code, you can switch perspectives without changing your underlying metadata. The chips reflect how you want to look at the project right now, not a permanent decoration.

The broader idea is consistent with how ADL treats classifications generally: they are metadata you own, used both by the platform to inform what you see, and by templates to inform what is generated.

Try it out in your own project at https://app.agnosticdatalabs.com 🔗.