Writing
Blog
4 posts tagged “LLM”
Claude is Dreaming: How Anthropic Borrowed a Billion Years of Evolution to Fix AI Memory
Anthropic just shipped a feature called Dreaming for Claude agents. I had read the exact same idea three weeks ago in a book about the human brain. Here is what is actually happening and why it matters.
Stop Wasting 80% of Your LLM Tokens: The Caveman + Graphify Framework
A three-step framework — compressed prompting, knowledge graph indexing, and project-specific shorthand — that cuts token usage by 80-90% without sacrificing technical accuracy.
ModelRouter: Intelligent Multi-LLM Routing to Stop Burning Claude Pro Tokens
How I built a local proxy that classifies prompts and routes them to the right model — Gemma for trivial tasks, Gemini for features, Codex for debugging, Claude only for hard architecture problems.
The Context Gap: Why Frontier LLMs Fail at Local Business Problems
Frontier LLMs are trained on Western-centric internet data and fail at regional supply chains, local business etiquette, and jurisdiction-specific law — here's how localized fine-tuning fixes this on commodity hardware.