NEMO: AI Assisted Deskbot
Abstract
NEMO is an intelligent desktop assistant that performs tasks via voice commands in natural language with an average response time of 1.2-1.8 seconds and a command accuracy of 92-95%. It features a wake-word detector, real-time speech-to-text transcription using Google's Speech API, and a dual-mode natural language understanding (NLU) pipeline that leverages both rule-based matching and the Gemini API for generative AI. This enables NEMO to manage both structured and unstructured natural language inputs. In terms of safety, NEMO implements permission-based execution, white-listing of applications, and validation of user inputs to provide secure execution. NEMO streamlines desktop operations and tasks, facilitating a 30-45% reduction in time taken to complete tasks, increasing hands-free productivity. NEMO employs a distributed computing architecture, with a laptop as the main processing unit, and it is scalable for future integration with other devices (such as Raspberry Pi) and network protocols.