LICENSE: QNFO Content License Agreement v1.1 (https://github.com/QNFO/license/). The agentic AI research paradigm enables solo investigators to match or exceed the throughput of traditional research teams. We survey 16 government, academic, and private-sector programs deploying autonomous AI agents for scientific discovery, identify five recurring architectural patterns, and present quantitative benchmarks demonstrating that the complexity gradient from single-domain to multi-domain operation is shallow (1.7x token cost for 4x domains, near-flat wall-clock time). Building on the QWAV/QNFO research program (9 Zenodo publications, 25x-90x speedups in preliminary self-experiments), we design a unified multi-domain architecture spanning protein design, atmospheric chemistry, climate physics, and power grid topology.