What We Built

The National Archives and Records Administration (NARA) faced an overwhelming backlog of military personnel record requests that threatened veterans' access to critical benefits. NARA's National Personnel Record Center (NPRC) serves as the central repository for millions of military and civil service personnel documents.

Many of these records remain paper-based, severely limiting remote processing capabilities during the pandemic. By 2023, the backlog of Official Military Personnel File requests alone had grown to 530,000. Manual processing allowed staff to handle only 30-50 records per full workday, with processing timeframes stretching into months.

Veterans seeking time-sensitive benefits eligibility verification faced significant delays. Limited staff resources were stretched thin by growing request volumes, while manual processes created inconsistent results and accuracy challenges in record location and retrieval.

Working with NARA and SoftRams, Fearless implemented robotic process automation solutions that transformed how the agency processes veterans' records requests.

With thirteen two-week sprints to complete this work, we conducted discovery alongside development. Our Scrum approach drove iterative progress, promoting continuous collaboration with our customer. Two to three times each week, we conducted knowledge transfer sessions with NARA staff, allowing our engineers to document each bot's functionality in detail.

Our solution dramatically improved processing capabilities. During one demonstration, two bots processed 25 records in just 30 minutes with 90-95% accuracy - a task that would typically take a full working day for staff to complete manually. The final bots deliver consistent processing times of just 1-3 minutes per record, dramatically outpacing manual approaches.

To ensure reliability, we built sophisticated exception handling mechanisms that provide fail-safes when errors occur. The system captures error reasons and automatically sends notifications to appropriate staff. This balance of automation with appropriate human oversight ensures quality while maintaining the efficiency benefits of automation.

Outcomes

90–95%
Accuracy achieved in record location and retrieval.
1–3
Minute average processing time per record.
30-50
Previous manual records processed per full workday.