All Posts · Thinking in Tokens

2026

Test-Time Compute Explained: Why Sampling More Beats Training Bigger, and When It Doesn't Fri 25 September 2026

The Anatomy of an Agent Loop: Why One LLM Call Is Not Enough Thu 24 September 2026

Agent Memory and Skills: What to Keep, What to Load, What to Forget Thu 24 September 2026

Long-Context and State Management for AI Agents Wed 23 September 2026

How AI Agents Turn Text Into Actions: Tool Calling, JSON, CodeAct, and MCP Tue 22 September 2026

AI Agents Forget Instructions Because Context Compaction Deletes Them Sat 19 September 2026

LLM Agent Costs Are Quadratic: The Token Trap and the Prompt Caching Fix Sat 19 September 2026

What Is an LLM Agent? Inside the Policy Engine and the Harness Loop Thu 17 September 2026

What Is an AI Agent and How Do You Build Your First One ? Sun 13 September 2026

Welcome to Thinking in Tokens Wed 09 September 2026