Page 241 of 506 · Back to latest writing
Customer debt turns visible revenue into hidden operating drag when bad-fit deals, custom promises, and unresolved value gaps burden delivery. Distinguishing repeatable learning from unsupported obligations helps leaders decide which growth is worth its full cost.
Narrative debt emerges when a company’s official story lags operating reality, leaving teams to act from conflicting maps. A reset names what changed, what remains true, which assumptions must retire, and the behavior that should follow from the updated story.
Coordination debt turns informal speed into recurring delay as companies grow, because cross-functional handoffs still rely on relationships and custom negotiation. Repaying it means designing clear interfaces with owners, inputs, decision rights, service levels, and escalation paths.
Trust debt accumulates when leadership’s words repeatedly diverge from observable reality, prompting people to wait, hedge, and seek private interpretations. Repair requires naming the break, the affected audience, a specific new commitment, and proof points that make changed behavior visible.
Decision debt compounds when choices are deferred, undocumented, or left to decay, forcing the same tradeoffs through repeated debate and local workarounds. A ledger makes ownership, default behavior, ambiguity costs, and evidence thresholds explicit so stalled questions can move.
Debt accumulates across a company whenever speed, ambiguity, or convenience leaves future costs unnamed. A shared register makes those liabilities visible, clarifies their carrying costs and owners, and lets leaders choose what to repay, refinance, or consciously carry.
Brendan McCord, a former Navy submarine officer, wrote the Department of Defense’s first AI strategy and founded the Joint Artificial Intelligence Center and Cosmos Institute. His philosopher-builders idea joins classical philosophy with decentralized technology to preserve individual judgment.
Jerry Liu co-founded and leads LlamaIndex, an open-source framework connecting language models to custom data. His experience highlights models as reasoning engines, the unglamorous necessity of data parsing, and the realities of moving AI prototypes into production.