Engineering leader Jimmy Malhan says AI adoption exposes an organization’s culture in a way established technology never did. Mature tools arrive with precedent, documented practice, and people who already know the answer, and a team can lean on that expertise even when its internal communication is poor. AI offers none of it. The decisions are new, the evidence is thin, and the only mechanism available for evaluating a tool is the quality of the argument a team can have about it.
Malhan has built distributed systems and led teams through complex product and technical work, and he put a direct question to his own team early. “Are you ready for the upcoming changes?” he asked. He has spent the time since removing three habits that were preventing his organization from answering that kind of question honestly.
Making Disagreement Cheap
The first pattern Malhan identifies is silent disagreement. During a training platform migration, meetings closed with apparent consensus while the real debate continued in Slack direct messages for weeks. The plan looked settled, but it was not. “I changed the room,” he says. He began asking for objections aloud, by name, before any decision closed, and stayed hands-on while both the team and the problem were still new, deliberately lowering the cost of dissent until raising a concern became ordinary rather than confrontational.
Malhan is candid about what the alternative signals. “A quiet room is days away from your hardworking engineers polishing resumes,” he says, “and wearing a polite face.” For an organization evaluating unfamiliar technology, the objection that never reaches the table is the one that would have caught the flaw.
Naming the Owner in Writing
The second pattern cost his organization the most. Malhan calls it credit fog. “Work shipped and nobody knew whose call it was,” he says. Leaders tend to assume good work speaks for itself, and in practice credit drifts toward team leads and visible senior engineers regardless of where the decision originated.
His fix was administrative and took roughly 30 seconds. He recorded each decision in writing with four fields: the decision, the owner, the date, and a summary of the reasoning. The effect showed up quickly. “People stopped defending their turf once their name was safe,” he says.
Correcting in the Same Week
The third pattern is delayed feedback. Issues that waited for review season aged into resentment, and Malhan puts the arithmetic in blunt terms. “A two-minute conversation becomes a 40-minute conversation in six months,” he says. He moved feedback into the same week an issue appeared and stayed specific about it. He also varied the delivery, praising publicly and correcting privately in some situations and reversing it in others so people knew what was coming. He began with a seven-day cycle and loosened it as trust developed between teams. A correction delivered while a decision is still fresh costs almost nothing, since speed is the operative variable.
Culture gets built in small moments, and what registers is how people remember a meeting rather than what was announced during it. Left alone, these three patterns produce a recognizable sequence: disengagement first, then resignations nobody anticipated, then exit interviews that describe problems a leader could have addressed in the second month.
Building an organization ready for AI, with security and compliance intact, is a choice “rewarded by consistency and patience,” he says, rather than by any single initiative. To learn more, connect with Jimmy Malhan on LinkedIn.
