Ideas

Frameworks and mental models for building better systems and making better decisions.

Chunking Strategies for RAG

How you split documents determines RAG quality. Learn the five chunking strategies and when to use each one.

RAGArchitectureBest Practices
Explore framework →

Context Window Engineering

The context window is your most expensive and limited resource. Learn to treat it like premium real estate—every token should earn its place.

LLMArchitecturePerformance
Explore framework →

Permission Passthrough Pattern

When building AI systems that access data, never bypass existing permission systems. Pass user context through every layer to ensure AI sees only what the user is allowed to see.

SecurityEnterprise AIArchitecture
Explore framework →

Progressive Complexity Principle

Build systems that start simple and allow complexity only when needed. A design philosophy for creating software that scales with requirements, not ahead of them.

System DesignArchitectureSimplicity
Explore framework →

The Agent Reliability Paradox

The more autonomous you make an AI agent, the less reliable it becomes. Here's how to navigate this fundamental tension in agent design.

AI AgentsSystem DesignReliability
Explore framework →

The Eval-First Principle

Build your evaluation harness before building your AI system. Without measurement, you're optimizing for vibes—and vibes don't scale.

AI EngineeringEvaluationBest Practices
Explore framework →

The Three Lenses Framework

A mental model for evaluating technical decisions through three critical perspectives—user impact, team velocity, and system health.

Decision MakingArchitectureLeadership
Explore framework →