Strategy, pipeline, budget, portfolio and resources are one economy, not five processes. The business side asks for investment, the delivery side produces it, money and people are scarce, and every choice has an opportunity cost: the best option left out. Inside the firm there is a market, but it trades in influence, so its prices stay invisible.

Grounded in Robbins on scarcity; Stein, and Scharfstein and Stein, on internal capital markets

Connects to Idea 01 · Idea 02

What the firm values, measured rather than asserted. Priorities come from structured comparison, not typed weights. Economics asks three things of them: any two can be compared, the comparisons hang together, and more is better only until a target is met. Some investments have no substitute; others are alternative ways to solve the same problem.

Grounded in Saaty; Keeney and Raiffa

Connects to Finding 5 · Idea 01

The proposals are a demand for capital; read the other way, they are the firm's supply of opportunities, and no selection can fund a priority that nothing in the pipeline serves. Three sources feed it: proposals from the business, which cluster on familiar work; a top-down search opened where the portfolio says a priority is starved, in the manner of the theory of constraints; and candidates derived from the lifecycle of the firm's assets as they age. The pipeline sets the ceiling; the optimizer only decides how close the firm gets.

Grounded in Penrose on the firm's productive opportunity; Goldratt on constraints

Connects to Finding 2 · Note 03 · Idea 02

The budget, capacity, mandates, floors and splits are limits on the best portfolio, and each has a price: what the portfolio gives up because the limit is there. Only the binding limits matter. Because projects are funded whole, the price moves in steps rather than smoothly.

Grounded in Kantorovich, Koopmans and Dantzig on duality and shadow prices

Connects to Finding 3 · Idea 01

Internal and external people become delivered work. In the short run the internal workforce is fixed and only the mix and the external top-up move; in the long run everything can change. Output per person rises with specialization in an area of the enterprise and falls with the cost of coordination, and a fixed internal core directs the rest. Project-centric versus product-centric is short run versus long run.

Grounded in Coase; Penrose; Brooks

Connects to Idea 01 · Idea 04

Money and capacity are two currencies, and the firm converts between them at rates that differ by direction and by horizon. Money buys capacity quickly through contractors and services, and slowly but durably through hiring and training. Capacity converts back to money quickly only for contractors; internal capacity cannot be sold back in the short run, so idle internal capacity is money already spent. The conversion is asymmetric, like a currency with a wide spread, and that asymmetry decides when to top up, when to hold, and why idle capacity and external premiums carry a cost.

Grounded in Coase and Williamson on make or buy; Dixit and Pindyck on costly reversibility

Connects to Idea 01

Which capabilities to build, and how large. A capability is a shared fixed cost that many projects reuse, so judging projects one by one under-invests in it. Returns rise with specialization and fall with coordination; the right size sits between.

Grounded in Panzar and Willig on economies of scope; Penrose

Connects to Idea 04

Innovation moves the frontier. A better idea, a new business model or a different way to reach the same outcome delivers more strategy for the same budget. The prices show where a better opportunity is worth most: the priorities that cost the most to leave short.

Grounded in Schumpeter; Solow on productivity

Connects to Finding 2 · Note 03 · Idea 02

Run-Grow-Transform, Three Horizons and pace layers split the budget in advance, by percentage. The discipline tags each investment by what the money does, maintain, deliver, build or retire, and lets the split come out of the optimization. A floor such as build at least 20% enters as a limit with a visible price, and Horizons and pace layers become views of the result rather than inputs to it.

Grounded in Baghai, Coley and White on Three Horizons; Ramsey on allocation over time

Connects to Idea 01 · Idea 04

How the budget is divided across business units, programs, capabilities or product lines before any investment is chosen: envelopes, splits and minimum allocations. For each budget the question is what more money would buy there, and money should move to where it buys most, unless a minimum or a floor holds it in place; then that rule shows its price.

Grounded in The equimarginal principle; Weingartner on capital rationing

Connects to Finding 3 · Idea 01

Which whole investments are funded within each budget. Sorting by score and cutting at the budget falls short of the best set and drops the large strategic bet, because the problem is an integer one; must-haves enter as priced commitments and alternatives as options that exclude each other. Selection runs in two modes. In the classic mode, projects are selected against money and people for the planning cycle. In the agile mode, epics are selected against each value stream's team capacity, increment by increment, on the same value scale.

Grounded in Dantzig and Gomory on integer programming; Weingartner

Connects to Finding 1 · Note 02

Every decision object moves through three states. Planning commits; managing corrects the course within the cycle; learning returns twice, to managing and to the next plan. Each state has its own decisions and owners, and under uncertainty each carries a tolerance as well as a value. Learning is what narrows the tolerance.

Grounded in Argyris and Schön on double-loop learning; Deming

Connects to Idea 04

A path, not a year. Run versus build is consuming versus investing; delay erodes value; cost and capacity are spread over periods. The best portfolio chosen each year alone can end behind the best path chosen across years.

Grounded in Dean and Hirshleifer on capital budgeting over time; Ramsey

Connects to Idea 04

First what uncertainty is: risk with known odds, uncertainty without them, and the unknowns nobody foresaw; and why expected values and risk heatmaps ask for probabilities nobody has. Then the turn: instead of predicting the uncertainty, ask how far each estimate, weight or limit must move before the decision changes. Wide margins are settled; thin ones go to people.

Grounded in Knight on risk and uncertainty; Box on models; Ben-Haim on info-gap decisions

Connects to Finding 4 · Note 01 · Note 04 · Idea 03

People decide with limited attention, and the system must save it in both directions. In eliciting: structured comparisons instead of typed weights, never a probability, only whether a number could be off by as much as the decision can tolerate. In understanding: a few prices and a margin instead of dashboards, explanations in plain words, and review sent only where the margin is thin.

Grounded in Simon on bounded rationality and attention; Cyert and March; Sweller on cognitive load

Connects to Idea 03 · Idea 05

People reject what a machine imposes and accept what they can adjust; they will use an imperfect algorithm if they can modify its output, even slightly. So every machine output can be accepted, modified or rejected, and every override is allowed and shows its price.

Grounded in Dietvorst, Simmons and Massey on algorithm aversion

Connects to Idea 05

Proposals arrive optimistic, ratings drift upward over cycles, duplicates hide in the register, budgets become mental accounts, and sunk costs keep projects alive. The discipline expects this and designs around it instead of assuming it away.

Grounded in Kahneman and Tversky; Kahneman and Lovallo on the planning fallacy; Thaler on mental accounts

Connects to Finding 5

Dominant business units, a delivery function with a natural monopoly, information held by one side, and influence in place of prices. Bounded rationality together with opportunism is why organizations need governance at all.

Grounded in Williamson on transaction costs; Jensen and Meckling on agency; Akerlof on adverse selection; Stein on influence

Connects to Idea 02

Rules, mandates, floors and authority each carry a cost. The discipline separates the rules that protect from the rules that are only obeyed, prices each, and keeps the record a board or an auditor asks for: who decided, on what, with what margin.

Grounded in Berle and Means; Jolls, Sunstein and Thaler on behavioral law and economics

Connects to Finding 3 · Finding 4

Three kinds of work in every decision. The optimizer computes the best answer for what is given, with its prices and margins, exactly and repeatably. AI draws out what people know, explains and flags: useful and fallible. People commit. Across the roughly 120 decisions in the chain, the strategy end is almost all judgment and the capacity end runs mostly on rules. A decision can be automated as far as its rule is explicit and its result reversible.

Grounded in Ricardo on comparative advantage; Simon

Connects to Finding 6 · Idea 05

What AI does: elicits what people know, explains what the optimizer proposes, drafts the asks and the record, flags what looks wrong. What it must not do: invent the numbers the optimizer uses, substitute a plausible answer for an exact one, or commit. AI also changes production itself: as it raises the productivity of delivery, the bottleneck moves to judgment and to the business's capacity to absorb change.

Grounded in Simon on attention; Agrawal, Gans and Goldfarb on prediction and judgment

Connects to Idea 05

Augmentation: people decide, and the machine informs, computes and explains. Supervised: the machine proposes, and people review where the margin is thin. Autopilot: the machine prepares the decision within rules, and people are called in when a rule trips; it prepares, it never commits. The right mode belongs to the decision, set by how explicit its rule is, how reversible its result and how wide its margin.

Grounded in Sheridan and Verplank; Parasuraman, Sheridan and Wickens on levels of automation

Connects to Finding 6 · Idea 05

Modes should not be fixed. A decision earns its way toward autopilot as its record proves reliable and its margin stays wide, and returns to people when the margin thins or results drift. That movement is how trust is earned rather than assumed. Looking ahead: which decisions are likely to move as AI improves, and which should never move because accountability must stay with a person.

Grounded in Lee and See on trust in automation; Bainbridge on the ironies of automation

Connects to Idea 05

The capstone. One function owns the whole investment system, strategy, pipeline, budget, portfolio and resources, across the cycle: what it decides itself, what it delegates to the containers, what it leaves to the machine, and how it answers to the board.

Grounded in Chandler on strategy and structure; Bower on the resource allocation process

Connects to Idea 01 · Idea 05