Weaviate Primer

Why Weaviate

Weaviate bundles vector storage, ANN indexing, and an optional embedding module into a single Go binary. No sidecar model server, no Redis queue for indexing — just one process that speaks GraphQL, REST, and gRPC.

Core Concepts

• Class — analogous to a table. Defines the schema, vectorizer, and index config.
• Object — a row. Holds properties, a vector, and an optional cross-reference.
• Vectorizer — the module that turns text into vectors at insert time (e.g. text2vec-transformers).
• Hybrid search — BM25 keyword + vector similarity fused into one ranked result set.

Multi-Tenant Setup

Use a single Weaviate instance with per-tenant classes (e.g. TenantA_Document) or rely on the native multi-tenancy feature (v1.20+) that partitions objects under one class by tenant key. The latter keeps schema maintenance trivial.

Quick Start

Nimbus Integration

Point the Nimbus loader at your Weaviate endpoint for runtime telemetry clustering. Store session fingerprints as vectors, query nearest neighbors to detect license-sharing rings, and surface anomalies via the dashboard's real-time feed.