As the number of disasters increases, the first responders still face challenges in carrying out their activities of rescuing and maintaining the people’s safety in a disaster affected area. In disaster frontline operation under the disruption of telephone/Internet services, information handling like collecting, sharing and reporting disaster information takes time, that is a hindrance to their activities. First responders usually rely on verbal and manual approaches to the information handling. It causes difficulty in the accurate and timely information distribution that forces the first responders to operate with the limited information. To meet this challenge, we propose a Front-Line Operation System (FLOS) that enables first responders to efficiently collect and share information at the disaster sites. FLOS is a portable ICT system comprising of small server, local communication equipment like Wi-Fi and battery. The server acts as a standalone Web server which locally offers social networking service functions including chat, bulletin board and voice/video communications. First-responders are able to access to the system through their smartphones/tablets and use it to locally collect and share necessary information even under telephone/Internet service disruption. FLOS also provides AI (artificial intelligence) features like speech recognition, intelligent image processing and automatic analysis such as locating objects from photos, acquisition of information from external systems. This on-site analysis can be timely used to support the first responder’s activities. With FLOS, it will be possible to quickly grasp and analyse situations on site in any communication environment. It is capable of semi-automatically acquiring, visualizing, sharing, reporting, and recording of disaster information and the system utilizes the features for rescue/search/recovery activities which makes the first responder activities be hands-free. In this paper, FLOS concept is proposed and its required specifications for the edge computing to include AI functions are discussed through a preliminary experiment.