Building the Use Case Portfolio

The ten-minute review

The senior team sat down with the season's shared document. Nine validated use cases on the page, each a paragraph long. The review took ten minutes. Two graduated into Solutions. One rerouted after the flag caught something the scoring had missed. One was parked until the relevant policy work completed in the next quarter. Five stayed in the sandbox for another cycle under tightened conditions.

That afternoon, the board meeting had AI on the agenda. The conversation took forty minutes. The senior leader walked the board through the nine use cases, the verdicts, the reroutings, and the reasoning. The board did not need to litigate any of it. The portfolio was the conversation.

A year earlier, the equivalent board conversation had taken close to four hours and produced no decisions. The difference was not the board. It was that the senior team finally had something to show.

What a portfolio contains

The portfolio is a short document. Not a deck. Not a roadmap. A single shared page that holds, for each validated use case, six things.

The pattern it sits under, from the eight-pattern catalog. The scoring verdict, across the four dimensions. The flag verdict, from the ethical and relational review. The boundaries inherited from Safety. The named owner, who will be accountable for what happens next. And a one-paragraph description specific enough that someone reading the portfolio cold could understand what the use case actually is.

That is the whole format. Everything else is detail that belongs in the experiment record, not the portfolio. A portfolio that has grown into a multi-page brief has stopped being a portfolio and become a report.

The shape of a good portfolio

The portfolio has a size. Five to ten validated use cases is the working range.

Fewer than five usually means the sandbox did not do enough work or was too narrow in scope. A three-use-case portfolio is not a portfolio; it is a pilot. More than ten usually means the team failed to reject candidates that should have been rerouted or parked. A fifteen-use-case portfolio is almost always a backlog wearing a portfolio's clothes.

Inside the range, the portfolio has a shape.

Two or three use cases are high value, low risk. These scored well on the four dimensions, cleared the flag, and fit inside Safety's boundaries without stretching them. These are the first use cases that graduate into Solutions. They are the ones the senior leader will point to when the board asks what AI is actually doing for the organization.

One or two use cases are high value, high risk, held with caution. These scored well on some dimensions and raised honest concerns on others. Usually they raised a flag that was not red enough to reject but yellow enough to matter. These do not graduate into Solutions yet. They move into the Skills stage as the kind of work senior staff are being formed to hold responsibly, and they may graduate into Solutions later, under stricter conditions.

The remainder are parked. Not killed. Not live. Waiting on a changed condition. Sometimes the change is a policy update Safety is working on. Sometimes it is a capability the organization does not yet have. Sometimes it is simply that the team did not have the formation to hold this particular use case responsibly yet, and the parked status is an honest admission rather than a postponement.

Use cases that were rejected outright do not appear in the portfolio. They appear in a separate rejection log, kept because the next team proposing something similar should be able to see what happened the last time, and for what reason.

What the portfolio is not

A roadmap. The portfolio is not a promise about when anything will ship. It is a snapshot of what the organization currently has enough evidence to stand behind.

A priority stack. The portfolio does not rank use cases against each other by expected return. The scoring produced the confidence; the flag produced the legitimacy; the portfolio is the set of things that cleared both bars. Among the cleared set, priority is a separate leadership conversation that happens after the portfolio exists, not during its construction.

A launch calendar. Graduations from the portfolio into Solutions happen on a timeline the Solutions stage owns. The portfolio's role ends when the handoff happens; it is not a project plan.

The portfolio is a diagnostic snapshot of what the organization has learned and what it is willing to stand behind. Treating it as anything more makes it brittle. Treating it as anything less wastes the work that produced it.

Why size discipline matters

Portfolios above ten use cases are almost always mislabeled backlogs. Real organizational capacity at this stage is small. An organization that has just finished its first sandbox season does not have the skilled staff, the formed judgment, or the governance maturity to responsibly hold twenty simultaneously active AI use cases. Pretending otherwise front-loads failure.

Size discipline is a forcing function. Five to ten means the senior team has to cut. They have to say no to candidates that passed the scoring and cleared the flag but are not among the most important things the organization should be doing with AI this season. This cut is unpopular with the staff who championed the rejected candidates, and it is the right cut.

The discipline works because scarcity produces honesty. A portfolio with unlimited slots attracts false confidence. A portfolio with ten slots, and a queue, attracts the actual work of discernment.

Who owns the portfolio

Senior leadership. The portfolio is a leadership artifact, not a project artifact. If the senior leader cannot talk through the portfolio in ten minutes, each use case, why it is there, what risk it carries, what it will unlock, it is not a portfolio. It is a spreadsheet.

This is a specific accountability, not a figurehead role. The senior leader should know the content of each use case, not just the summary. They should be able to answer a board question about any specific use case without consulting notes. They should be willing to defend any of them in front of the people the use case affects. A senior leader who cannot meet this bar has delegated the portfolio too far down; the fix is not a smaller portfolio but a senior leader who has done the work to own it.

The experiment teams, the peer reviewers, and the flag writer all contribute. The portfolio, as the artifact the organization stands behind, belongs to the senior leader.

What the portfolio unlocks

Three things, in order.

Skills training gets organized around what graduated. Teams learn on the specific workflows the organization already validated, not on generic tool fluency that may or may not apply to the work. This turns Skills as Formation, Not Training from an aspiration into a curriculum.

Solutions deployments get scoped to the high-value, low-risk use cases. The procurement conversation that follows is narrower, faster, and better grounded, because the organization already knows what it is buying tools to do. This connects to canon #17, Why Solutions Come Last: with the portfolio in hand, Solutions stops being a shopping trip and becomes a deployment against validated demand.

Board conversations about AI become portfolio conversations. The right level for AI to sit on the board's agenda is the portfolio: what graduated, what was rerouted, what was parked, and what the senior team is watching next season. This is the only kind of AI conversation mature organizations should be having with their boards, and the portfolio is what makes it possible.

Closing the season

With a portfolio in hand, the Sandbox stage closes its first season. The organization is ready to move up the staircase. To Skills, where the validated use cases shape formation. And eventually to Solutions, where they cross into live work under trained, governed hands.

The next season of the sandbox reopens when the portfolio needs refreshing, which is usually a quarter or two later. Sandboxes are not a one-time stage. They are a recurring rhythm that keeps the organization's AI posture aligned with the work the organization is actually doing, which changes faster than any static portfolio can keep up with.

The portfolio is a diagnostic snapshot of what the organization has learned and what it is willing to stand behind.