Business imagination is the scarce asset now

What the paper develops

You chair the portfolio review that decides which AI ideas receive scarce capacity. One team brings a working assistant. Another has a vendor demo. A third shows time saved on drafting and analysis. All three can produce something. None of that evidence tells you which work deserves to change, what result would matter, or whether AI is the right investment.

Cheaper execution is useful. Models can draft, summarize, classify, compare, search, synthesize, and code quickly. A field study of customer-support agents found a 14 percent average productivity gain in one defined workflow, with larger gains among novice and lower-skilled workers. At enterprise level, McKinsey reported broad regular AI use alongside much lower reported EBIT impact. A defined task can improve while the portfolio still lacks a basis for choosing among investments.

That gap has a cost. Capacity follows the most visible evidence: the polished prototype, the familiar vendor, or the metric that is easiest to count. A promising operating problem may remain hidden because no one has framed it as a comparable decision. Another team may carry a demo forward without showing which business result it can change.

The first constraint is the judgment to frame the work

Portfolio investment requires evidence about the work and its consequence. Counts of users, prompts, generated documents, accepted code suggestions, and launched pilots can rise while cycle time, quality, customer experience, risk, and financial performance remain unchanged. McKinsey's five-layer measurement framework traces a value claim through technical performance, user adoption, operational indicators, strategic outcomes, and financial impact. Each layer answers a different question, and no single activity measure carries the whole investment case.

Research on abandoned AI projects points to the same upstream difficulty. Gartner cited poor data quality, inadequate risk controls, escalating costs, and unclear business value. RAND found that misunderstanding project intent and purpose caused more failures than any other factor in its study and advised leaders to focus on the problem before the technology.

Because cheap execution creates more candidates than the review can frame and compare, the first limiting asset is the judgment to turn one operating problem into a testable choice. This paper calls that judgment business imagination. It is the practical ability to notice where a workflow, decision, control, product experience, or customer interaction could work differently, state why the change matters, and make the possibility testable against other demands for capacity.

Put one opportunity brief before the tool decision

Require one opportunity brief before a material AI idea receives funding, scarce delivery capacity, or permission to enter recurring work. The brief names six things: the work or decision that should change; the known baseline; the mechanism by which value could appear; the evidence that earns another commitment; the accountable owner and the decision boundary that remains human-owned; and the routes that deserve comparison.

Consider a proposal to improve portfolio reporting with AI. That phrase is too loose to fund. A useful brief might reveal a narrower problem: analysts spend time reconciling inconsistent fields before an executive review, and the late reconciliation delays the decision. The value hypothesis is shorter preparation time and a more stable evidence pack. The first routes to compare may be data validation and a deterministic rule.

Compare routes before committing to a tool

The available routes include workflow redesign, a frontier or lower-cost model, deterministic automation, better data, a vendor capability, human expertise, training, a policy decision, waiting, and stopping. A short model exploration may fit a high-value question whose answer is uncertain. Stable, repetitive, low-risk work may fit deterministic automation. A data repair may resolve the reporting example without a model. The portfolio assigns people, models, data, automation, governance attention, and funding according to value, uncertainty, risk, repeatability, and evidence maturity.

Connect the pilot to the business result

Keep the opportunity brief attached after approval. The pilot should show whether the system met task-specific acceptance criteria, whether intended users put it into the workflow, whether the named baseline moved, whether that change mattered to the business outcome, and whether the result justified the full cost and risk. Indirect value can remain visible as a leading indicator without being converted into a financial claim before the evidence supports it.

At the next review, bring the same brief back with the evidence filled in. The portfolio can then explore, redirect, fund the next bounded commitment, integrate, wait, or stop. The three opening teams face one standard: name the work worth changing, the value at stake, the evidence needed, the accountable owner, and the route that fits the job before funding the tool.