Archive • Sgnt.ai • Buttondown

Understanding modern AI is understanding embeddings

For years I’ve wanted to write the “perfect” article on embeddings, to give anyone with enough maths to write simple Excel formulas a deep and intuitive...

When Users Won’t Wait: Engineering Killable LLM Responses

Standard LLM interaction models often block user input during generation, but what happens when that's impossible? We faced this exact challenge with an LLM-...

sgnt.ai: In-memory free-text search is a super-power for LLMs

I build LLM-based non-player characters (NPCs) for an online game. I've seen considerable improvement in a number of places by augmenting inference (token...