Compositional Learning-And-Reasoning for AI Complex Systems Engineering (CLARA) Information Session

The Defense Advanced Research Projects Agency (DARPA) Defense Sciences Office (DSO) has issued a Special Notice announcing an Information Session webcast for the Compositional Learning-And-Reasoning for AI Complex Systems Engineering (CLARA) Disruption Opportunity (DO). The session is scheduled for February 19, 2026, from 1:00 p.m. to 4:00 p.m. Eastern Time and will provide potential proposers from industry, academia, and government with comprehensive information about the CLARA DO vision, objectives, and mechanics. Advance registration is required and must be completed by February 11, 2026, at noon or when capacity is reached (350 individuals maximum). The Information Session is voluntary and not required to submit a proposal. DARPA anticipates releasing the full CLARA DO solicitation in February 2026 on sam.gov, with all resulting awards structured as Other Transactions (OTs) for prototype projects under 10 U.S.C. 4022.

The CLARA program is a fundamental exploratory research initiative designed to create high assurance, broadly applicable AI systems of systems through hierarchical composition of Machine Learning (ML) and Automated Reasoning (AR) subsystems, emphasizing verifiability, strong explainability through automated logical proofs, and hierarchical vetted logic building blocks. Eligible proposers should review Section 7 of the Disruptioneering Program Announcement (DARPA-PA-25-07) for specific eligibility information and requirements. Key deadlines include webcast registration closure on February 11, 2026, at noon, submission of one-page collaboration profiles by February 23, 2026, at 4:00 p.m. to CLARA@darpa.mil, and non-U.S. citizen security clearance documentation also due by February 11, 2026. Technical inquiries should be directed to Benjamin Grosof, Program Manager, at CLARA@darpa.mil. Participants should note that webcast recording is not permitted and no reimbursement is provided for participation costs.