Show HN: AI watched my screen for a year. Weather beat sleep (donethat.ai)

In a fascinating self-experiment, the creator of DoneThat tracked their screen activity for a year, generating insights into productivity through an AI that transformed screenshots into descriptive data. Using an LLM, they categorized 31,698 activity intervals, revealing surprising patterns about their work habits and productivity. Notably, they discovered that their median productive workday only amounted to 6.8 hours, challenging the conventional notion of the eight-hour workday. They also identified that weather and morning productivity were significant factors, with cold days correlating strongly with productivity and early mornings yielding the best focus. This project holds critical significance for the AI/ML community as it demonstrates the potential for AI to provide personalized insights that encourage better time management and workflow strategies. Additionally, it reflects the emerging trend of using technology to monitor and improve personal productivity, showcasing how behavioral changes can be triggered simply by the awareness of being analyzed. The experiment underscores the importance of structured workdays, effective goal setting, and the viability of AI as a supportive tool for self-improvement in professional settings.