Cookbook
Use-case driven guides for turning embeddings, tickets, events, feedback, and operational records into working EigenLake workloads.
Use filtered nearest-neighbour search, cursor pagination, and search-unit economics to retrieve relevant records from a corpus of tens of thousands of consumer complaints.
Discover natural groupings in large vector datasets with DBSCAN and k-means, from protein sequences to support tickets, without exporting data to a separate analytics stack.
Surface unusual sensor events, operational records, and outlier folds with Local Outlier Factor over vector embeddings — without pre-defining what normal looks like.
Discover emergent themes in multilingual text, scientific literature, or operational notes without manual labeling — using spherical k-means, c-TF-IDF, and optional LLM-generated labels.
Compare two time windows and find semantic shifts — emerging themes, growing clusters, and disappearing patterns — without manual trend analysis or external BI tools.
Chain search, clustering, anomaly detection, topic modeling, and temporal shift into one unified analytical pipeline — proving that vector data is a compute substrate, not just a retrieval index.