Module 06 Planned
Scaling Agent Systems
Prompt caches, semantic caches, model gateways, vector indexes, work queues — the agent CDN stack.
In this module
01Prompt caching as distributed caching, with new failure modes
02Semantic versus exact caches and cache invalidation
03Agent gateways and proxies (LiteLLM, OpenRouter, internal gateways)
04Vector indexes as distributed retrieval
05Work queues for asynchronous agent tasks
Lessons publish here as they are written.