Our interactions with built environments are increasingly augmented with digital capabilities. Smart appliances coupled with Building Management Systems (BMS) promise to increase occupant comfort and to reduce energy consumption through automated control and personalized services. However, it is not clear how smart appliances, originally designed for smart homes, can be used in the context of non-residential buildings. For example: How can a user be given access to smart appliances in her vicinity? This simple question raises a number of issues in non-residential buildings: What are the smart appliances in the vicinity of a person (this requires building-wide metadata collection and maintenance)? How can a user access new smart appliances as she moves around without a cumbersome initialisation process, yet with security measures enforced? How to mediate between diverging smart appliances settings? In this thesis, we study how Mark Weiser's original vision for ubiquitous computing can help us tackle such questions. Specifically, our contributions are the following: (i) At the human-building interface, we design and implement a system for a building manager that enables building-wide visualization and control by transforming the physical building, with its structure, sensors and actuators, into a virtual reality system. For building occupants, we enable the ubiquitous use of appliances and sensors in vicinity by bridging existing off-the-shelf smart appliances to a common Bluetooth Low Energy (BLE) interface with room-level based authorization. (ii) We explore different identification mechanisms to support metadata management. For existing BMS sensors and actuators, we introduce a crowd-sourced approach that incrementally builds up consistent metadata. For smart appliances in vicinity, we introduce acoustic based localization. We further design and implement a system that uses smartphone sensor based user feedback to automatically select appliance settings. We also mediate conflicts between users locally and building wide energy policies. (iii) On a system level, we present an architecture for the decentralized integration of smart appliances into non-residential buildings that relies on user smartphones as opportunistic gateways and BLE for communication. We design and implement a distributed framework to evaluate BLE performance in such smartphone-peripheral systems. We perform a detailed evaluation of multiple smartphone models that shows that the native BLE stack fails to provide homogeneous abstractions for different implementations. We improve the default behavior, by the introduction of a dynamic, smartphone model dependent library, that adapts to the idiosyncrasies of specific BLE implementations.