Tag
Tags are the fine-grained annotation layer. Each tag belongs to a tag set, each signal can carry many tags, and the same tag can be shared across signals to surface cross-cutting themes.
Tag sets
A workspace can have several tag sets — each a closed vocabulary for one dimension. A playbook seeds the workspace with starter sets (geographies, sectors, maturity bands) and you can add or edit sets at any time.
How tags are applied
The pipeline assigns tags during generate and merge based on the signal’s content and the workspace’s tag sets. You can also add or remove tags by hand on any signal — manual edits are preserved across re-runs.
Tags vs. categories
Categories are mutually exclusive: one signal, one category. Tags are not — a single signal can carry many tags, and the same tag can appear across many signals. Use categories to structure the corpus, tags to slice it.