How Enterprises Build AI That Lasts.
Models are increasingly shared. The competitive ground has moved — to the context layer the organization builds around the model, the proof-of-value discipline that distinguishes capability from value, and the operating-model change required to absorb either.
"A proof of concept shows that the technology can perform.
A proof of value shows that the work improves."
— Law 3
Three editions · One argument
Executive Edition
$19.99
paperback · $9.99 Kindle
The strategic spine in ~250 pages. Front matter, the Twelve Laws, Parts I, III, and VI of the full book, plus the Epilogue and Glossary. Built for executives, board members, and governance leaders deciding whether the AI investment is producing value.
Pre-order on Kindle →Full Edition
$39.99
paperback · $19.99 Kindle
All 650 pages. Strategic foundation, data and governance, system-building (LLMs, prompts, RAG, agents, multi-agent), operating chapters (tooling, security, cost), value, adoption, future. The reference manual for the people who will build, govern, and operate the program.
Pre-order on Kindle →Reader Edition
Free
with newsletter signup
The full book as a polished PDF designed to read on screen — cover prepended, chapters opening on recto pages. The version reviewers and corporate evaluators get. Available on launch.
Notify me when ready →Kindle pre-orders open today; ebook ships Summer 2026. Paperback releases on the same day; join the list to be notified the moment paperback is buyable on Amazon.
The framework
The argument of the book, compressed into twelve operating principles. Short enough to quote in a board meeting. Substantive enough to hold the program accountable to them.
Models are the smallest part of enterprise AI advantage.
The Context Layer is the durable asset. Everything else depreciates.
A proof of concept shows the tech can perform. A proof of value shows the work improves.
Baseline before you build. A weak baseline is more useful than no baseline.
Hallucinations are a structural property of next-token prediction, not a bug.
A disclaimer is not a control boundary.
Bounded autonomy: permissions, reversibility, and escalation are first-class design decisions.
The evaluation harness is the precondition for everything else.
Three words for value: projected, realized, banked.
Every AI system needs a written rule under which it will be taken offline.
Authority should match accuracy.
Operating-model change is not the consequence of AI adoption. It is the price.
Working artifacts
Three downloads adapted from real engagements.
Self-Assessment
Twelve questions, one per Law, scored 1–5. Produces a maturity readout in four bands. Use for self-diagnosis, cross-functional comparison, or initiative-specific review.
Download (with newsletter signup) →Charter
The operating contract between a proof-of-value team and the executive sponsor: baseline, value hypothesis, target users, evaluation method, eval owner, adoption criteria, risk controls, scale-or-stop rule. Fillable template plus the two Framework Quality Tests.
Download (with newsletter signup) →Diagnostic
Twenty questions across seven sections assessing whether the organization has a named, governed context layer. Maturity readout in four bands; the section minimum matters more than the total.
Download (with newsletter signup) →Sample chapter
Chapter 6 — Data Governance in the AI Era — opens with the MyCity, MyAI public-sector AI failure: a chatbot that operated for 28 months under official city branding while answering legally consequential questions wrong. The chapter develops the AI Data Trust Model and the structural controls that distinguish "we deployed AI" from "we govern AI."
Read Chapter 6 (free with email) →
Author
CEO of Modali Consulting, where his work focuses on helping enterprises adopt AI in ways that compound value rather than capability.
Faculty at the Cornell Brooks School of Public Policy. Formerly Chief Technology Officer at the New York City Mayor's Office of Contract Services, where he led major digital-transformation work in procurement and public-sector technology.
Holds AI-related patents in cost modeling and procurement. Authored the MyCity, MyAI public-policy case study at Cornell — the public-sector AI deployment that frames the governance argument across Chapters 6, 16, and 19 of the book.
Speaking · Workshops · Advisory
The Twelve Laws Self-Assessment runs as a 90-minute leadership workshop. The Proof-of-Value Framework and the Context Layer Diagnostic anchor longer engagements. Modali Consulting works with enterprises on the operating disciplines this book develops — context-layer architecture, proof-of-value design, AI governance, and the operating-model change that turns AI capability into banked value.