# Retrieval Weaviate Collection Schema and Vectorizer Config
Status: public
Confidence: medium (0.685) (verified)
Last verified: 2026-06-03
Generation: ai_structured


## TL;DR

Weaviate collection schema and vectorizer configuration determine what gets vectorized, how vectors are named, and how tenants or shards are isolated.

## Core Explanation

Retrieval agents need to distinguish query relevance issues from collection-definition issues. A poor answer can come from the wrong property schema, missing vectorized field, mismatched named vector, tenant selection, or index settings rather than from the language model.

Before changing prompts or rerankers, an agent should capture the collection name, properties, vectorizer configuration, named vector configuration, multi-tenancy settings, replication and sharding configuration, and any migration needed to rebuild vectors for existing objects.

## Source-Mapped Facts

- Weaviate documentation describes collection creation as configuring collection parameters through a client library. ([source](https://docs.weaviate.io/weaviate/config-refs/collections))
- Weaviate documentation says that if a collection definition does not explicitly define a named vector, Weaviate creates a single vector collection. ([source](https://docs.weaviate.io/weaviate/config-refs/collections))
- Weaviate vectorizer configuration documentation says a collection can define multiple named vectors, each with its own vector index. ([source](https://docs.weaviate.io/weaviate/manage-collections/vector-config))

## Further Reading

- [Weaviate Collection Definition](https://docs.weaviate.io/weaviate/config-refs/collections)
- [Weaviate Vectorizer and Vector Index Config](https://docs.weaviate.io/weaviate/manage-collections/vector-config)