COST OPTIMIZATION TECHNIQUES

PRACTICE LAB CONNECTION In the hardening stage of the Resume Review Agent lab, which is Chapter 69, you will add model routing and prompt caching. Route simple field extraction, such as pulling a name and a list of job titles from a resume, to a cheaper tier. Keep the final narrative report, which needs judgment, on the stronger tier. Place the system prompt and tool definitions at the start so that they can be cached, and keep changing details at the end. Then use the prompts below against the lab, measuring cost before and after.