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

The Defense Advanced Research Projects Agency (DARPA) Defense Sciences Office is soliciting innovative basic and applied research concepts for the Compositional Learning-And-Reasoning for AI Complex Systems Engineering (CLARA) program, which focuses on creating a scientific, theory-driven architectural foundation for hierarchical composition of Machine Learning (ML) and Automated Reasoning (AR) subsystems with high assurance properties. The program comprises two technical areas: TA1 addresses the development of new high-assurance ML/AR composition approaches with explicit theoretical substantiation, logical explainability properties, and polynomial computational time complexity, while TA2 focuses on developing a software composition library that integrates validated TA1 approaches with standardized data formats and interfaces. Key contract requirements include demonstrating verifiability with logical explainability, supporting multiple AI kinds in composition, achieving polynomial computational complexity, and improving task reliability and sample complexity over state-of-the-art baselines; additionally, all software developed must be open-sourced under an Apache 2.0 or similar commercialization-friendly license, and Associate Contractor Agreements are required between TA1 and TA2 teams within three months of award. Proposals are due April 17, 2026, at 4:00 p.m. ET, with abstracts strongly encouraged by March 2, 2026, at 4:00 p.m. ET, and must be submitted electronically via the DARPA BAA Portal as zip files not exceeding 100 MB. All performers must participate in program-wide activities including Phase Kickoff Meetings, Workshops, and Hackathons, with evaluations based on the technical merit and feasibility of proposed approaches, team qualifications and prior integration expertise (particularly for TA2), and the clarity and completeness of risk mitigation strategies.

This opportunity is issued as an Other Transaction (OT) for Prototype project with no set-aside designation and no identified incumbent contractors, carrying a combined maximum award value of $2,000,000 for Phase 1 (Feasibility Study) and Phase 2 (Proof of Concept) combined, inclusive of government funding and any performer cost share, with Phase 1 limited to $1,350,000 and Phase 2 limited to $850,000, allowing up to $100,000 to be shifted between phases. Phase 1 will run for 15 months while Phase 2 will run for 9 months, with the anticipated program start date of June 22, 2026. The solicitation encourages proposals from diverse performer types including large businesses, small businesses, non-profits, academic institutions, and laboratories/FFRDCs, though FFRDCs and UARCs have specific eligibility restrictions requiring documentation or organizational conflict of interest mitigation plans. Performance of the research will take place at the contractor's facilities as specified in individual proposals, with work conducted in unclassified settings and application task domains that feature explicit state-of-the-art performance benchmarks. This Disruption Opportunity was originally posted February 10, 2026, and has been amended twice (March 11 and April 9, 2026) to clarify technical requirements, extend proposal deadlines, and specify complete attachment requirements for conforming proposals.