Work with data
Use the appropriate view for records, documents, embeddings, relationships, and time.
Minds presents several ways to inspect data inside an instance. These views answer different questions; they are not different plans.
| View | Use it for | Watch for |
|---|---|---|
| Key-Value | Retrieving or updating a record by a known key | Keyspace and value type |
| Document | Inspecting structured records | Actual fields and schema, not a sample row |
| Vector | Comparing embeddings in an index | Dimension, metric, embedding model, and index configuration |
| Graph | Entities and named relationships | Edge direction, provenance, and time scope |
| Time Series | Values ordered by time | Timestamp meaning, units, and aggregation window |
Start with one small dataset
Choose an authorized namespace and a clear keyspace name. Insert a small, non-sensitive example with the exact API schema. Read it back by identifier before adding indexing, search, or visualization. This keeps an ingestion error separate from a query error.
Use key-value storage, vectors, knowledge graph, and analytics for the registered operations. The UI's category names do not imply that a route exists at a similarly named path.
Check measurements at the source
Some dashboard views include placeholder values or fallbacks in the audited source. For counts, sizes, and latency used in decisions, confirm the corresponding live API result and its measurement period. A well-rendered panel is not proof that its numbers came from your Mind.