About $2.00M to automate correction of data-quality problems in the Army personnel and pay system — buying not the model but what sits beneath it (XPERNO)

1Buying what sits before the model

Spending on AI brings to mind buying a model or an analytical system. What this contract addresses is what comes before. Errors and inconsistencies accumulated in a personnel and pay system are found automatically, ranked for which to fix first, and corrected. As a premise for using AI, answers cannot be right where the data is not.

Here AI is used not on the side that produces answers but on the side that prepares the ground for producing them.

1. 1DetectionErrors and inconsistencies in the data are found automatically
2. 2TriageWhich bear on operations and which can wait is judged
3. 3CorrectionThe fixes are made in that order

2The choice of the word triage

The description places triage between detection and correction. The term comes from medicine, naming the decision of treatment order under limited resources. Finding data-quality problems does not mean all of them can be fixed. A step is needed to judge which bear on operations and which can wait. That the judgment itself falls within what is automated shows this contract is not aiming at a simple bulk correction.

3Personnel and pay as the subject

What is handled is the Army Integrated Personnel and Pay System — the machinery managing who holds which position and how much they are paid. Errors in that data surface as errors of assignment or of payment. It draws less attention than equipment or operations, yet it is an area where a mistake reaches a person directly.

The period runs two years, from April 2026 to April 2028, and at about $2.00 million it belongs to the majority: 190 of the 222 records this site holds as of 2026-08-31, or 86 percent, fall under $10 million.