ND Starts With Use Cases

ND is not a warning label. It is how you make large‑scale retrieval practical.

Use cases

• Search at scale (when exact is too slow)
• Recommendations (when small ranking drift is acceptable)
• Clustering and similarity exploration
• RAG retrieval (fast candidate generation)
• Anomaly detection (approximate neighbors at scale)

ND is safe when its bounds are explicit and monitored.