All posts
3 articles
Filter by tag
-
The Axe Came Down Again
We learned, painfully, that splitting a monolith along noun boundaries was an architectural mistake. Twelve years later, we're doing it again with agents.
-
Thinking Tokens and the Reasoning Gap in Agent Systems
How LLMs use thinking tokens to reason better within a single call — and why that reasoning silently vanishes the moment you build an agent loop.
-
How LLMs Actually Work: The Intuition Behind Every Token
A visual, ground-up guide to how large language models turn text into intelligence, from tokenization to attention to generation.