Use Batch API and/or flex processing for nightly/backlog sweeps, regression scanning, or "second opinion" on merged PRs. Outputs become training and eval data. The cheapest path to 100% coverage.
Async review outputs serve multiple downstream purposes, creating a continuous improvement flywheel.
Mode E complements real-time modes (A, B, C) by ensuring 100% coverage over time. Use it as the background safety net that catches what synchronous reviews miss.