WatchBOT aims to create a Physical Artificial Intelligence (PAI) system to detect, manage, and resolve
malfunctions in remote terminals and edge devices, enhancing reliability, safety, and user experience. Devices
like public car chargers and informational kiosks often suffer from malfunctions that require on-site hardware
resets, resulting in high operational costs, user dissatisfaction, and downtime. WatchBOT addresses these issues
by integrating IoT sensors, multi-agent AI, and explainable AI into a domain-scalable framework that enables
proactive, intelligent, and explainable system management. Central to WatchBOT is an AI engine that performs
real-time monitoring, anomaly detection and forecasting, and root cause analysis. A network of specialized multi-
agent systems, coordinated by a central LLM-based “mind,” enables flexible, dynamic collaboration and
autonomous resets. WatchBOT leverages the AutoGen framework for adaptive data handling across diverse input
types, making it suitable for various sectors. WatchBOT also enhances trust through user-friendly natural
language explanations (NLEs) and real-time voice interaction, guiding users and technicians through issue
resolution while explaining AI-driven decisions. Scalable and secure, the architecture supports both centralized
and edge deployment. The project aligns with the SID themes of Social Innovation and Advanced Digital
Technologies, promoting efficient urban management, sustainability, and inclusivity. WatchBOT’s design
emphasizes green AI practices and supports explainable, user-centered AI deployment.

Runtime: 2025 - 2028