Notes on retrieval infrastructure.

Practical writing about vector storage, search systems, and the decisions behind them.

August 2026

Unstructured data should become retrieval-ready on arrival

A serverless approach to moving files from raw input to useful search context without managing a processing fleet.Read article

August 2026

The cheapest vector storage in the world

Why storage, writes, and retrieval should be metered as work instead of reserved as infrastructure.Read article

August 2026

What is a vector database?

A direct explanation of embeddings, distance metrics, metadata filters, and retrieval.Read article

August 2026

Talqora vs Turbopuffer

Two object-storage-native approaches to vector retrieval, compared through operating model and control plane.Read article

August 2026

Talqora vs Cloudflare Vectorize

Where an edge-adjacent vector store fits and when an AWS-region index is a better operational choice.Read article

August 2026

Talqora vs Pinecone

Comparing a managed vector database with a retrieval layer that does not require standing capacity.Read article

August 2026

Talqora vs Vespa and Qdrant

When a self-operated search cluster is the right choice, and when an API-first retrieval layer wins.Read article