Overview

Layeron Vector (vector) gives your backend app a typed vector index backed by Cloudflare Vectorize in your Cloudflare account.

You declare the index once in application code. Layeron provisions the underlying Cloudflare Vectorize index in your account, wires the Worker binding, and routes runtime operations through the product worker.

With Vector, you can:

• Store vector embeddings with application-defined IDs and metadata.
• Run similarity search with cosine, euclidean, or dot-product metrics.
• Filter query results by indexed metadata fields.
• Read, update, and delete vector records by ID.
• Inspect index configuration at runtime.

Vector is in preview. Indexes, metrics, and dimensions are fixed at creation time.

Use Cases

Vector supports any workflow that needs embedding-based retrieval:

• Semantic search — Match query embeddings against document chunks.
• Recommendations — Find items with similar embeddings.
• Product matching — Compare product embeddings by category or feature.
• Future RAG — The foundation for retrieval-augmented generation pipelines.

Next Steps

• Get started: Declare an index, store vectors, run similarity search, and deploy.
• Indexes: Configure names, dimensions, metrics, bindings, and metadata indexes.
• Data: Insert, upsert, get, delete, describe, and batch vector records.
• Search: Query by similarity with limits, metadata filters, vector return values, and metadata return values.
• Limits: Plan around dimensions, batch sizes, metrics, metadata indexes, and immutable index settings.
• API reference: Review vector options, records, mutation results, query inputs, matches, and module methods.