norika1207-lab/alfred-butler

Alfred is a personal AI butler system built by and for one individual, designed to act as an always-available assistant across voice, text, and mobile interfaces. The project is primarily documented in Traditional Chinese and appears to be a sophisticated personal project rather than a general-purpose tool for public use. At its core, Alfred is a FastAPI Python backend with 147 API endpoints and 69 AI-powered tool functions, paired with a custom iOS app written in Swift. The system integrates with Google Calendar, Gmail, Google Drive, LINE messaging, Telegram, Twilio phone calls, HealthKit, and the device's GPS location. It includes scrapers for 13 Taiwanese and Korean shopping websites to compare prices, a voice bank of over 3,000 pre-recorded mp3 response clips, and a 309 MB SQLite database pre-populated with data about restaurants, hotels, travel itineraries, and other personal information. A key performance feature is a set of "fastpath" functions, 17 common request types like checking the time, recording attendance, searching files, or responding to greetings, that bypass the AI model entirely and respond in under one second. More complex requests go through AI reasoning. You would use or study this project if you are interested in how a complete personal AI assistant can be architected: combining a heavyweight language model for complex tasks with deterministic fast-paths, a pre-built voice response library, deep integration with productivity services, and an iOS front-end for voice interaction.