Research

Research references and guides behind the toolkit — how agent architectures hold up in production, what a graph has to guarantee before you query it, and how to pick a model from the job rather than the leaderboard.