Case Studies
DLRA's products are evaluated in controlled operational settings with partner defense agencies. The following case studies document system performance against manual baselines — measuring retrieval accuracy, analyst workflow time, and adoption patterns under realistic intelligence processing conditions.
Partner Agency Intelligence Brief Evaluation
A controlled evaluation with partner-agency analysts measured SynthBrief's impact on intelligence brief production workflow. Analysts using SynthBrief with sentence-level provenance produced signed-off briefs in an average of 47 minutes, compared to 4.2 hours for manual production — an 81% reduction in total workflow time. The evaluation also documented the adoption failure of polished end-to-end output and the design pivot to provenance-exposed generation.
Maritime Signals Analysis Pilot
A pilot deployment of Maritime NLP in a maritime surveillance context measured the system's ability to process maritime text data alongside AIS and sensor feeds. The pilot documented a 40% reduction in maritime threat report triage time and evaluated entity extraction accuracy across maritime communications, incident reports, and OSINT feeds.