In January 2010, Japan’s prime minister asked a retiree to take over the country’s flag carrier. Kazuo Inamori, who had founded Kyocera, would turn 78 at the end of the month. On January 19, Japan Airlines filed for bankruptcy protection. By the count on Inamori’s official site, its debt came to about 2.3 trillion yen, and his archive describes the staff he found as lacking a sense of crisis, or any feeling that the company was theirs.
He wanted to know what the old JAL had not tracked closely: whether each route made money. By his account, the profit of every route and every flight became available the following day. Each month, departments presented their figures over a three-day meeting, and while he listened he told their leaders what to do next.
Three days a month is a lot of a chairman’s calendar. Inamori thought it bought more than tidier accounts. The profit system, he told more than 4,000 business managers in 2012, “led to fundamental changes in the organizational climate.”
What the boss keeps asking about
Edgar Schein of MIT’s Sloan School of Management had described this move long before. In a 1983 paper on how founders create culture, he listed ten ways leaders get a group to try things their way, running from explicit teaching to messages “of which even the founder may be unaware.”
By 1988 he had sorted them. Five he called “primary embedding mechanisms”: what leaders pay attention to, measure and control; how they react to critical incidents and crises; deliberate role modeling and coaching; the criteria for handing out rewards and status; and the criteria for recruiting, selecting, promoting and retiring people, along with what he called “excommunication.” Charters, creeds, buildings and the org chart he filed as secondary, for “articulation and reinforcement.”
Teams learn what really counts from what leaders watch, react to and reward. Inamori’s meeting is the first mechanism at its plainest: a department whose numbers go before the chairman every month learns what he cares about without a memo.
The moments everyone watches
Of the ten, Schein wrote that in his observation the most potent were role modeling, what leaders pay attention to, and how they react to critical events. One founder in the 1983 paper, called C, chided in public any manager who overran a budget and did not tell him when the problem was first noticed, “no matter what the reason.” The lesson, in Schein’s words: “poor ultimate results could be excused, but not being on top of one’s situation could never be excused.” He taught it, Schein wrote, “through the manner in which he reacted to situations.”
Drawing on training groups, Schein later explained why charged moments teach fast: everyone witnesses them and tension is high, so what happens next “will very likely create a norm.” A learner driver takes in more from the second the instructor hits the dual brake than from a chapter of the manual.
What the two papers rest on
Both papers draw on Schein’s consulting and observation; neither uses a controlled design to show cause and effect. The 1983 paper rests on “a dozen or more entrepreneurs” he had observed, and in 1988 he granted that this clinical method lacks “the methodological rigor of controlled empirical work.” Embedding, he added, only gets a group to try certain responses, with “no guarantee” they will work.
At JAL, the credit Inamori gives his profit system is his own account, and in the same lecture he also credited his philosophy and “a Divine intervention.” The same years brought salaries down to around 20 percent below rival All Nippon Airways, pensions halved for current staff, 521.5 billion yen of debt forgiven by banks and 350 billion yen from a government-backed fund, according to the aviation expert Hajime Tozaki. Nobody can pull the meeting’s effect apart from all that.
The weekly review
Imagine a team lead who keeps telling her engineers to raise risks early. Her weekly review opens with one question: is each project on track? The engineer who admits a slipping date gets twenty minutes of questions in front of everyone. The one who quietly patches a slip gets nothing. A couple of months later, every status report says green.
Nothing she said was wrong. What she asked about, and what she reacted to, taught the team that bad news is expensive. If her first question became “what’s at risk this week?” and the first answer earned help instead of a grilling, the same meeting would teach the opposite. Westrum’s typology asks whether bad news can travel upward.
Counting the AI
AI adoption runs on the same machinery. Amazon has set a target for 80 percent of its developers to use AI each week, the Financial Times reported. A group of employees built “Kirorank,” a leaderboard ranking staff by their activity on Kiro, Amazon’s AI developer platform. According to the FT, as reported by heise online and HCAMag, some staff began assigning autonomous agents to needless tasks to climb it, padding their token consumption and pushing up costs.
It was switched off in late May 2026. Senior vice president Dave Treadwell told employees it had been built with “good intentions.” He also told them: “Please don’t use AI just for the sake of using AI.” Amazon said the dashboard “was not a formal or approved tool.” It now reportedly tracks “normalised deployments,” how regularly developers use AI to write useful code.
The limits are real: the account rests on one newspaper’s sources, the board was unofficial, nobody published how many people gamed it, and whether the new measure is harder to game is unknown. But the pattern is Schein’s first mechanism meeting Goodhart’s Law: people learned what was being counted, and produced the count.
An executive can say “don’t use AI just for the sake of using AI.” What gets asked about at the next review will show whether anyone meant it.