Start Here
Start with the AI system or operating-model problem you need to solve.
Choose by Decision
| If You Want To | Start Here |
|---|---|
| Build reliable agents | State Machine Pattern: make control flow explicit. Then: Agentic Contract, Supervisor Pattern, or all Patterns. |
| Debug agent failures | Probabilistic State Machine: trace failure through a pipeline. Then: Context Window Fallacy or Conservation of Chaos. |
| Evaluate agent behavior | Validator Asymmetry: check outputs before trusting them. Then: Golden Dataset or Jurisdictional Voids. |
| Choose model tiers | Intelligence Arbitrage: match model capacity to the task. Then: Iron Triangle or Utilization Paradox. |
| Design AI operating models | Judgment Compression: assign ownership of AI decisions. Then: Human-in-the-Loop or Jurisdictional Voids. |
How Arizen Is Organized
- Writings - essays and field notes as they are published.
- Concepts - recurring mechanisms for architecture and operating-model decisions.
- Patterns - implementation structures for production systems.
- Books - long-form synthesis across essays and patterns.
If your team needs an architecture decision for a system it depends on, start with Advisory.