Is Prompt Caching Actually Lowering Your AI API Bill?
Find out whether prompt caching cuts your AI API cost per usable result. Compare cache writes, reads, retries and review time, then check the provider's bill.
LLM operations
Your AI workflow has logs. Can they explain one bad decision?
Learn how to trace a bad AI decision across documents, model calls, tools and APIs, keep sensitive data out of logs, and test whether a fix actually works.
Your AI Model Has an Expiry Date: Build the Migration Lane Early
AI models and endpoints retire. Learn how to inventory dependencies, test replacements, run a controlled release and preserve a usable rollback route.
OpenAI Responses API: Migrating After the Assistants Shutdown
Move legacy OpenAI Assistants workflows to the Responses API with a practical plan for recovering history, choosing conversation state, testing tools, and controlling costs.
Written recommendations from Trafik og Veje, Aarhus Municipality (2011) and AgroTech (2010).
