Notes on context engines, structured memory for LLMs, and the decisions behind how NeuralGraph is built.
Graphs aren't powerful because they're complicated. They're powerful because they let incredibly complex behavior emerge from incredibly simple rules.
