Core Concepts

Claro is built around a small set of concepts that show up across every surface. Once you know these, the rest of the platform is consistent.

​

Catalogue

A typed, persistent object — Products, Suppliers, Line Items, Variants, etc. — that holds your records. Each catalogue has its own schema (Attributes), data (Records), upstream feeds (Data Sources), transformations (Operations), and governance (Config). Multiple catalogues coexist in a workspace and can reference each other (e.g. Product.supplier → Supplier).

​

Attribute

A typed field on a catalogue: text, number, image, date, enum, or reference. Attributes carry validation rules, required flags, enum value sets, and (for references) the related catalogue.

​

Record

A row in a catalogue. Each record has a stable identity, a value per attribute, and a per-field history. Inline edits create change events with provenance.

​

Data Source

An upstream input that populates records. Sources include file uploads, scheduled scrapes, HTTPS pulls, supplier portal submissions, and database connectors. Each source has its own mapping to the catalogue’s attributes and its own schedule.

​

Operation

A verb of the platform — Validate, Normalize, Bulk Enrich, Find Duplicates, Generate Taxonomy, Push & Sync, etc. Operations read records, optionally call models or external systems, and write back changes. Every operation produces a run with inputs, outputs, confidence per change, and a reversible diff.

​

Run

A single execution of an operation. Runs are auditable: you can see which records were read, what changed, what didn’t, and why.

​

Confidence

A 0–100 score attached to every model-driven write, derived from source reliability, content consistency, model certainty, and retrieval quality. Each operation has configurable auto-apply, review, and reject thresholds.

​

Provenance

The chain of custody on every value: the operation, run, data source, model, and reviewer that produced it. Roll-back is per-field.

​

Taxonomy

A first-class hierarchical, versioned object — not a free-text label. Taxonomies can be generated from your data, imported from a standard (GS1, Google, internal), or hand-edited. Taxonomy assignments are themselves tracked operations with confidence scores.

​

Similarity & Duplicate graph

A workspace-level graph across records using embedding similarity and rule-based matching. Find Duplicates, Find Similarities, supplier match, and variant grouping all read from this graph.

​

Knowledge Base

A workspace-scoped collection of documents (datasheets, brand guidelines, catalogs, policies) that operations can cite when generating or matching content.

​

Notification

A queued review item from an operation whose confidence fell below the auto-apply threshold. Notifications can be approved, rejected, or edited inline; bulk actions are available per cluster.

​

Credit

A unit of usage. Operations consume credits per record processed; cost varies by operation type and model selection.