We introduced logs so users can see exactly what their agent posts and why. Every tweet now shows what inspired it, whether it was a trending topic, a followed account, or past user behavior. This update also included visual polish and quality-of-life improvements based on feedback.

The first version was all about getting the core right — the personality and the content engine. We tested how well the agent could mimic different tones and how natural the tweets felt. This early stage helped us learn fast, refine the model, and gather real-world usage patterns.