Enterprise software has spent twenty years perfecting the art of telling you something is wrong. Almost none of it helps you fix it.
The notification is the end of the system’s responsibility. What happens next is the human’s problem.
The gap in practice
A supply chain analyst receives an alert: a shipment is delayed, a warehouse is at capacity, a constraint has been violated. The notification is accurate and timely. What happens next?
The analyst opens four systems. Downloads two CSVs. Cross-references three reports. Identifies the problem. Calculates a fix. Implements it manually. Logs the action. Files a report. Forty-five minutes for a problem that should take thirty seconds.
The gap between detection and resolution is not a technology problem. It is a design problem. The systems were built to observe, not to act.
Why the gap exists
The notification-to-execution gap exists because enterprise software was designed around data visibility, not decision support. The dashboard is the product. The analyst is expected to do the reasoning.
This made sense when the reasoning required domain expertise that systems could not replicate. It no longer makes sense when language models can understand context, constraint solvers can calculate optimal fixes, and the human’s job is to approve rather than compute.
Closing the gap
Closing the gap requires rethinking what the system’s job is. Not: detect and notify. But: detect, reason, propose, await approval.
The human still makes the decision. But the system has done the work to make that decision fast, informed, and reversible. The analyst’s job becomes one of judgement rather than computation.
That shift — from computation to judgement — is the design challenge of the next decade of enterprise software.