Abstract: A computer-implemented system controls peripheral devices using artificial intelligence. Audio or video data captured by one or more microphones or cameras is processed to determine a participant's desire or related contextual information. Based on the determination, one or more peripheral devices are caused to act, such as by presenting related content or control outputs. The system may further note user sentiment regarding the actions taken and update a deep reinforcement learning model, enabling inference of subsequent desires. Sentiment may be derived from the processed audio or video data. In some embodiments, the system presents accompanying or supplemental context, optionally displayed via different portions of a user interface, to improve user interaction and understanding.

Abstract: Systems and methods are disclosed for analyzing video data from one or more cameras to determine object association and room readiness within an environment. The system detects at least one person and at least one object, correlates the object to an identified person, and determines that the person has exited while the object remains. Based on the determination, one or more control actions are automatically initiated, such as sending a notification, updating a scheduling application, or presenting a visual alert. In another embodiment, video data is processed to detect readiness factors including, e.g., chair count, cleanliness, or whiteboard markings. A readiness score is compared to a stored threshold to determine whether the environment is ready for use, and control actions are initiated accordingly, including, e.g., reassigning meetings or adjusting building management settings.

Abstract: An intelligent audiovisual control system is provided that utilizes a hierarchical, agentic artificial-intelligence framework. A large language model (LLM) operating as a room agent interprets input data and delegates execution to one or more task-specific LLMs that perform specialized functions. The system establishes communication among agents and includes an environment-assignment module for binding agents to physical environments or functional domains, and defines supervisory and subordinate relationships across physical environments. Each agent monitors sensor data and actuates connected control systems such as lighting, HVAC, and audiovisual equipment. The architecture enables scalable, context-aware, and privacy-preserving coordination of intelligent environments through distributed reasoning and autonomous task delegation.

Abstract: An intercom system is provided having a plurality of intercom stations coupled to a digital network, each station transmitting intercom data onto a channel of the network. Each station receives intercom data from each other station and selects channels to be amplified and output to a speaker. Each station may listen to and talk to any other station or any selected group of stations. Processing of signals from the network is performed in each station, thereby eliminating the need for a centralized mixing and switching device. Stations may be added merely by connecting them to the network. The intercom system may be configured in a non-blocking mode or in a blocking mode.