Keen Bean ships macOS app that transcribes local audio and generates draft specs without call bots

Keen Bean launched as a native macOS application that captures system audio during meetings to generate structured outputs. It creates task cards, decisions, and draft specifications in Markdown format while the conversation occurs. The application avoids any call participation, relying solely on the host device's microphone input.

Data from the app description indicates processing occurs through external transcription services but routes content directly without intermediate servers owned by the developer. Files remain as local Markdown and JSON, integrable with tools like Obsidian. This setup addresses privacy concerns in regulated environments where bots are prohibited.

Analysis shows the tool targets users who manage their own documentation workflows, such as consultants handling client NDAs. By starting drafts during discussions, it reduces post-meeting writing time compared to pure transcription services. The subscription model provisions API access internally, eliminating user management of keys.

Future developments may include expanded model integrations or cross-platform versions, though current focus remains on Apple Silicon and Intel Macs.