Prompt caching cut our LLM bill by 60%
How restructuring prompts to put static content first let us take advantage of provider-side caching with almost no code change.
Jordan Lee
AI Correspondent
How restructuring prompts to put static content first let us take advantage of provider-side caching with almost no code change.
Details
This is placeholder seed content for local review — replace with real reporting before publishing.
More from AI Tools
We tried four vector databases so you don't have to
A practical comparison of Pinecone, Weaviate, Qdrant, and pgvector for a mid-size RAG pipeline, with real latency numbers.
Our eval stack, six months in
What we got right and wrong building an internal LLM evaluation harness, and why we eventually adopted an off-the-shelf tool for regression testing.
Picking an agent framework in 2026: a field report
LangGraph, CrewAI, and a hand-rolled state machine — tradeoffs we found only after shipping each to production.