A portfolio of funded AI initiatives is not yet a plan. In this stage of the capstone you decide when the enterprise does what it has already chosen to do, and what each phase has to prove before the next one is allowed to start. Working from a readiness pack that separates a workflow problem from a data problem, you place every initiative you funded into three fixed phases, land a quick win first, and name for each phase what it grounds and what its evidence releases. You choose one operating-model decision per phase — who owns delivery, what the work runs on, how money is released, who is made able — and you state the investment by phase and the incremental benefit by year. The output is the second section of your AI Transformation Executive Brief: a one-page roadmap and operating model a board could fund phase by phase, against gates carrying the authority to stop.

A portfolio with owners nobody named and systems nobody authorized is not governed, however many principles it endorses. In this stage of the capstone you decide who is accountable for each AI system and for the program itself, what each system may decide without a person in the path, and which five risks a board is asked to carry. Working from two evidence packs — a hiring screen that cannot reconstruct its own contribution to a decision, and an operations agent taking thousands of actions a week and asking to take more — you set a decision-rights setting for every system touching consequential decisions, separate the autonomy a system was authorized to have from the autonomy its records can actually support, and build a five-row risk register in which every mitigation names a condition somebody could later test rather than an activity somebody could merely run. The output is the third section of your AI Transformation Executive Brief.

A program with a roadmap and a governance framework still has to say what it is worth. In this stage of the capstone you decide how the change is made and how the value is proven — the point at which the brief stops describing intent and starts committing to numbers a board can hold you to. Working from an adoption survey, a set of documented metric baselines and a generated ROI workbook, you finish a change plan around the management layer that has to act differently, classify every benefit line as hard ROI, soft ROI or strategic option value with the reason attached, and name the assumption that closes each gap between what a sponsor claims and what is observed today. You then set a KPI for every bet, each carrying a baseline and targets for both years. The output is the fourth section of your AI Transformation Executive Brief: a one-page change and value plan in which every figure can be checked.

Four good sections are not yet a brief. In this stage of the capstone you assemble the use-case, roadmap, governance and value sections into one AI Transformation Executive Brief and find out, for the first time, whether they were ever really one argument. You place each section into the master template slot for slot, gather and deduplicate the appendices behind it, and then run an integration check by category rather than by section — candidate names, phase names, owners, KPI names and figures, each with a recorded result — repairing every mismatch in the section that carries it rather than in the summary. You confirm that each of the five questions leadership could not answer has a place in the brief a reader can be pointed to. Then, last, you write the one page that opens with the answer: what the enterprise can document today, what you propose, what you are not proposing, the risks you are carrying, and what you need from the board.