Tokens = Money. Most people burn 3–5× more than they need on Claude without realizing it. I put together a 2-page cheat sheet covering everything I wish someone had handed me when I started: Page 1 — the essentials - Pricing per 1M tokens (Opus, Sonnet, Haiku) - The Savings Stack (these multiply, not add — 80–95% cuts are normal) - The 5 core tactics: right-sizing the model, prompt caching, shorter prompts, output control, context trimming - Common token wasters to stop doing today Page 2 — the deep dive - Claude Code in the terminal: slash commands, CLI flags, how to slim your CLAUDE.md - Prompt caching full Python example (what to cache, what the response tells you) - Prefill tricks that kill the "Sure, here you go..." preamble - Model decision tree (when Haiku, Sonnet, Opus each) - Real-world recipes: cheap classifier, long-doc Q&A, cost-capped agent loops, batch evals - A pre-ship checklist you can run through before deploying 📎 Attached: Token_Saving_Cheat_Sheet Drop a comment with the one tip that surprised you most, or your own token-saving trick. Vincent

▎ EDIT (4 May): Amazon terminated my KDP account on Saturday. Book is now at https://payhip.com/b/VpTj3 — see [this post](link to your new update) for the story. I shipped a 305-page book this week. "Rewiring Corporate Finance", a blueprint for what the finance function becomes when AI moves from pilot to load-bearing. 12 chapters, 9 custom diagrams, EU AI Act compliance map and a 12-month execution plan for CFOs When I say I used Claude, people assume it wrote the book. It didn’t. I have over 15 years inside finance at Philips and Vodafone.The seat and the judgment are mine. What Claude actually did was this: 1. Pressure-tested the thinking: Every chapter had a thesis. I made Claude argue against it before I wrote.Some ideas held. Others didn’t. 2. Cut the corporate voice: Anything that sounded like a consulting deck got flagged.The writing is tighter because of that. 3. Handled the messy production layer: Cover math. Spine width. EPUB. KDP formatting. Claude wrote the Python to turn the draft into something uploadable.That’s where most people get stuck. 4. Turned ideas into diagrams : I described the systems. It generated clean visuals I could actually use. What it didn’t do: The experience.The opinions.The calls on what goes agent-first vs stays human. If you’ve ever thought about writing a book: This is way more doable now than it was even a year ago. If you want, I can break down exactly how I went from idea to published. Just DM me. And if you’re building AI skills on top of your current finance role: Comment on this post and I’ll send you the PDF of Chapter 9. It covers: - the four literacies (finance, data, AI, programming) - the hybrid finance-developer profile - why the junior pipeline is going to change