The AI-Native Enterprise: Three Transformation Rounds, One Twelve-Month Map: Running case study: Meridian Re, a reinsurance company Eight source-code demos included — Node.js, LLM - Tapa blanda

Sinopsis

Adding AI to a company does not make it an AI company.

Most companies have bolted AI onto the outside of an unchanged process: a chatbot on the website, a "summarise" button, a copilot in the inbox. The three-day task still takes three days. This book gives you one test to see where you really stand — take the AI out, what is left? — and a twelve-month map to get from "we have AI" to an enterprise that operates AI-Native: the machine reads, extracts, checks and proposes; people approve at clearly defined gates; take the AI out and the machine stops.

It is a build manual, not a survey. One company, Meridian Re, runs through every chapter, from the night it loses a contract to a rival that quotes in two hours, to the day its board approves a company-wide programme. Three rounds, each standing on the evidence of the last:

• Round 1 — one piece of software. Rebuild one real application in five layers: intent-based interface, RAG memory, tool calling, a three-agent pipeline with generative UI, and guardrails — PII masking, fallback, cost meter, an acceptance test set and blind grading before go-live.


• Round 2 — one department. Redraw the process with the 4P framework, build the department workboard where the AI does 80% before a person sees the file, run the two-session training plan, and replace input KPIs with value measures.


• Round 3 — the whole company. Answer the board's hardest question (if competitors can do the same, where are we ahead?), form an AI task force that replicates instead of doing the work, unify the data into a warehouse the AI can query safely, and open an internal AI gateway where any employee builds an assistant without code.

Every chapter opens with what you will be able to do and closes with a self-test. Every chapter ships a running demo: nine Node.js applications, nothing to install, working with Gemini, Claude or OpenAI through one configuration file — printed in full in the appendices and available to download. Deterministic parts come with tests; the book is honest about what a machine can verify and what a person must grade blind.

You also get a complete board-paper template with a milestone-released budget, a one-page AI policy in six articles, a two-session workshop plan, and a fifty-point checklist to score your own company every quarter.

Written for the executive who must fund the change, the manager who must run the department afterwards, and the engineer who must build it — in plain language, with no metaphors, and a command to type on Monday morning.

"Sinopsis" puede pertenecer a otra edición de este libro.