Tag
A self-improving AI framework that simultaneously updates both model weights and task-specific agent architecture via a language-model feedback agent, achieving significant gains across legal classification, GPU optimization, and biological denoising tasks.