← Labs On-device · no API key
AI Lab · Data-Quality Hub

Three ways embeddings guard your data.

One on-device embedding model, three jobs rules can't do: infer what a schema's fields actually mean, measure semantic drift between two data snapshots, and tell breaking schema changes from safe ones (with rename detection). All in your browser.

Embedding model loads on first run (~25 MB, cached after).

Schema Whisperer — infer structure & meaning from examples

Drift Lens — semantic drift between two snapshots

Breaking-Change Radar — schema diff with rename detection