Art of Novel Signals: Predicting and Forecasting with High Confidence
SBIR Opportunity Analysis
The Department of Defense, through DARPA, is seeking Direct to Phase II SBIR proposals to develop and demonstrate a predictive geopolitical forecasting model that utilizes temporal knowledge graphs and in-context learning. Performers must build a forecasting system that harvests untapped, multilingual open-source signals—such as radio, local news, and community audio—from data-sparse regions to extend event-forecasting timelines from days to weeks. The scope of work includes establishing a radio data engine, fine-tuning automatic speech recognition for predominantly oral languages, implementing a synthetic data strategy, and integrating paralinguistic audio feature extraction. Testing and evaluation will focus on critical strategic regions of interest, including Central and Southeast Asia, East and Northeast Africa, and South America. Proposals are due by August 19, 2026, and Phase II includes an 18-month base period followed by a $500,000, six-month option period for operational transition and rapid language onboarding.