Free Lesson
Cut AI Agent Costs with Prompt Caching
60 min
Nov 12, 2026 12:00 PM
By continuing, you agree to Maven's Terms and Privacy Policy.
What you'll learn
Measure what each agent run costs
Count input, cached and output tokens per step and per agent.
Lay out prompts so the cache hits
Put stable tools and instructions first so repeat calls bill at the cached rate.
Pick the right model for each step
Send simple steps to a smaller model and keep the large one where quality needs it.
Why this topic matters
Agents resend the same instructions, tools and history on every step, and you pay for them every time. Most teams notice only when the invoice arrives, with no idea which agent or prompt caused it. Live, I measure one agent's spend per run, reorder its prompt so the stable parts hit Claude's prompt cache, and move the simple steps to a cheaper model. Part 4 of 7 in the Agentic AI Systems series.
