Didon is an AI powered work and time tracking application that automatically generates daily work logs and journals. The app runs in the background on a Mac computer, capturing a screenshot every 30 seconds to determine how a user is spending their time. Instead of requiring manual input, it uses a local large language model (LLM), specifically Qwen-3-VL:2b, to analyze the captured screen data directly on the device. This analysis extracts a structured work journal, logs tasks, and categorizes activities by project and context. All screen analysis happens locally, meaning screenshots and sensitive data never leave the computer. The application offers several features for tracking and understanding work habits. It can detect specific windows, file names, applications, and tasks being worked on. Users can provide project goals and activity categories as context, which the AI uses to intelligently categorize work and deliver relevant insights. The app provides an activity breakdown, time spent on projects, and allows users to ask natural language questions about their work logs. Logs can be exported to CSV for client billing, internal reporting, or personal productivity analysis. The app starts automatically with the system and pauses when the Mac sleeps to avoid counting idle time. Typical use cases include professionals who need to bill clients by the hour, teams that want internal reporting on how time is spent, or individuals seeking to improve their productivity through detailed daily logs. The app requires a macOS system and operates with a one time payment model that includes lifetime access and updates. No additional AI model costs are incurred because the analysis uses the local LLM. Installation is limited to one system. Upcoming features are planned, such as integrations with other tools, advanced reports, the ability to add notes, an AI productivity coach, and work personality type analysis. A Windows version is on a waitlist for early access.