Blog

• October 30, 2026News · ResearchIntroducing Causal ReflectionWe are releasing initial benchmark results for a new model system architecture that enables companies to own their intelligence without fine-tuning, post-training, reinforcement learning, or routing.
• October 27, 2026Learn · ProductWhat is an Agent Loop?A language model can't remember or act on its own. The agent loop is the code that gives it both.