KALKIGEN JOURNAL

Journal notes on
AI

Practical thinking on engineering, intelligent systems, spatial products, and the craft behind digital work that lasts.

Latest thinking

Vector databases, explained for product teams

How embeddings, vector indexes and similarity search work together, and how to design a vector database layer that keeps AI answers fast, relevant and private.

What makes an AI answer trustworthy?

A field guide to retrieval, evidence, and useful safeguards for private knowledge assistants that teams can rely on.