Idler builds reinforcement-learning environments for model evaluation and training. Off the shelf or customized, focused on coding, reasoning, and economically valuable knowledge work.
