Blog

Writing on Token Management.

Context windows are a budget. These posts cover how I keep raw intermediate output out of the main session, pick the right model for the job, and get more done per token without babysitting the run.

• Jul 2026 Using Anthropic Fable Efficiently Fable is Anthropic's new frontier tier, and it's astonishingly good. The fastest way to burn through it is to let it write the code. The better way is to treat it like a principal engineer. Token Management
• Jul 2026 Managing Your Context Window with Agents Parallel subagents aren't just faster. They keep raw intermediate output out of your main context window. Here's how I structured my health data warehouse around that idea. Token Management