The Playbook Is on Amazon

Three months ago, the question every IT Director was asking was: how do I explain this Claude invoice to my CFO?

The question every CFO was asking was: why does this line item keep growing and why can nobody tell me what it's buying us?

We built PromptKing to answer both questions at the platform level. But a platform is only half the answer. The other half is the practitioner knowing what questions to ask in the first place — what the billing models actually do, where the waste hides, and why 80% of enterprise finance teams are guessing their AI spend within a margin wider than ±25%.

That's why we wrote the book.

What's in It

AI FinOps: The Complete Playbook — Third Edition is the field guide for IT Directors, CFOs, and FinOps practitioners managing enterprise AI spend across multiple vendors. Seven chapters, seven vendor playbooks, four appendices — including a 30-term glossary that now covers every concept the platform surfaces.

The third edition adds three chapters that didn't exist in the second:

Salesforce AgentForce — the full vendor playbook: Flex Credit billing, Digital Wallet blind spots, the three questions Digital Wallet cannot answer, EU AI Act Annex III risk tiers, and exactly how to govern an AgentForce deployment alongside Claude, Copilot, and Bedrock in a single view. Salesforce Summer '26 moves multi-agent orchestration to GA on June 15. This chapter lands at the right moment.

Open-Source and Self-Hosted AI — the fastest-growing AI cost category that generates zero vendor invoices. How to apply the same seat utilization and cost-per-task discipline to Llama, Groq, and on-premises inference that you apply to SaaS vendors. The absence of a bill does not mean the absence of cost.

How AI Cost Data Gets to Your Platform — the telemetry collection chapter. Push vs. pull collection, OpenTelemetry GenAI convergence, vendor coverage tiers, and why data freshness is a governance decision. Before you can see, optimize, or govern your AI spend, data has to flow. This chapter explains how.

The Board View Architecture

The second edition introduced Board View panels — a structured, typographically distinct block at the opening of every chapter designed for CFOs and board members who need the strategic picture without the practitioner depth.

Every Board View follows six elements: The Situation, The Number, The Risk, The Decision, What To Do Monday Morning, and the PromptKing feature that answers it.

One book. Two reading paths. Every audience.

What It Covers

The seven vendor playbooks in the third edition:

• Anthropic Claude — subscription plans, programmatic billing split, prompt caching, batch API, the June 15 inflection point
• Microsoft Copilot + Studio — M365 seat utilization, Copilot Studio agent billing at $0.01/message, GitHub AI Credits 27× multiplier
• Google Gemini — Workspace bundling, Flash vs. Pro 10× price gap, Vertex AI dynamic pricing
• AWS Bedrock — three billing modes, provisioned throughput waste, Bedrock vs. native Anthropic API decision framework
• GitHub Copilot — AI Credits transition, model multiplier table, org-level policy recommendations
• IBM Watsonx — Resource Units, governance scoring, bias and drift monitoring, regulated industry requirements
• Salesforce AgentForce — Flex Credits, Digital Wallet gaps, EU AI Act Annex III, multi-agent cost governance

Plus Part One (FinOps fundamentals, the five universal metrics, 90-day practice plan) and four appendices including the AI Subscription FinOps Terminology Reference.

Available Now

$24.99 paperback on Amazon.

The Kindle edition is in review — available within 24 hours.

Get it on Amazon →

---

The five seat archetypes — Ghost Seat, Underutilized, Normal, Power User, At-Risk — referenced throughout the book are patent pending with the USPTO (Application No. 2-01025019). PromptKing publishes this terminology as working vocabulary for the AI FinOps field.