The Materials Lab review did not go three hours. It went ninety minutes exactly, because Dahl watched the clock the entire time and stood up the second it hit the mark, whether or not anyone was mid-sentence.Kai had expected resistance. He hadn’t expected it to be this quiet, which was somehow worse. Dahl sat with his arms crossed through most of it, not interrupting, not objecting out loud, just watching Kai run the questions with an expression that made it clear he was cataloguing every minute as evidence for a case he intended to make later, to someone who wasn’t in the room yet.The moment that mattered came about forty minutes in. Kai had asked, in the open, unloaded way Constance had trained into him, how a recent handoff between materials testing and the design team had actually gone, not how the report said it went.A younger researcher, one of the newer hires, answered honestly before he’d thought better of it. “It went fine. It just took longer than it used to.”Dahl cut in

Six weeks into the pilot, the first real data set landed on Richard’s desk on a Tuesday morning, and by Tuesday afternoon he’d asked Kai to come by.Kai went in expecting resistance and found something stranger, a kind of quiet enthusiasm that put him on guard faster than resistance would have.“These are good numbers,” Richard said, sliding the report back across the desk like he wanted Kai to look at it again with him. “Fewer repeated errors in Archives. Cleaner handoffs in the Lab, even with your friend Dahl complaining the whole time. And this part here.” He tapped a line near the bottom. “Downstream time saved. That’s the number nobody was measuring for, and it’s the number that actually pays for itself.”“I noticed that too.”“How replicable is it.”“What do you mean.”Richard leaned back, the particular posture of a man about to ask a question he’d already decided he wanted the answer to. “If we wanted to run this in a third department. A fourth. Could it be documented well eno

Kai was walking out of Archives after a review, still half thinking about a note he needed to send Elena, when someone fell into step beside him without any of the usual throat clearing people did before approaching him.“Can I ask you something.”He looked over. He didn’t recognize her, though something about the pace she matched him at suggested she’d been waiting for the review to end specifically to catch him in the hallway rather than interrupting the room.“Go ahead.”“Is the root review actually meant to catch mistakes, or is it meant to teach people to notice their own mistakes before anyone else has to?”It wasn’t the question he was used to getting. Most people asked him about time, about metrics, about how to defend the process to a boss who thought it was slowing things down. This was a question about the mechanism itself, asked plainly, by someone who’d clearly already spent real time thinking about the difference.“Both, I’d say. Though I’d want the second one to matter

The pilot report had gone through three versions by the time Kai finally sent it, and even then he wasn’t sure the third was better than the first, only that it was different, and Constance kept telling him the language needed to land a certain way and hadn’t yet told him it had.“You’re describing what the process does,” she’d said that afternoon, reading over his second draft. “You need to describe what it costs and what it buys. Boards don’t fund descriptions. They fund trades.”He’d gone back to his desk and started again, and the second version had taken longer than the first, and by the time he looked up it was past seven and he’d missed dinner for the third time that week without meaning to, the way he kept not meaning to, each time telling himself this was the last late night before things settled.He got home a little after eight to a house gone quiet in the specific way houses go quiet once a child has been put to bed without his help. Lila was at the kitchen table with two

He spent three days on the draft, longer than he’d spent on anything since the original proposal, because this one had to survive contact with an entire institute instead of two departments willing to give him the benefit of the doubt.He built it the way he actually believed in it. Mentorship pairing at the center, senior staff matched with junior staff across a minimum term long enough to actually mean something. In person site visits, not surveys. A timeline that stretched to nearly a year before the full institute would be running the process the way Archives and the Materials Lab already were. He knew, writing it, that a year would sound slow to a board that had just used the word demonstrable in the same sentence as grant renewal. He wrote it that way anyway, because he’d rather defend a plan he believed in than apologize later for one he didn’t.He sent it to Richard on a Tuesday morning, framed as a budget sanity check, nothing more, a request to make sure the numbers held up