n this paper, we present the full reference implementation and field experiments of our real-time multi-agent adaptive tactical pervasive service collaboration (TAPS) platform. TAPS embodies our experience and lessons from our previous work on our agent- based pervasive service architecture (MAPS) (1) and the adaptive peer-to-peer agent (APPA) (2) software framework. We discuss the technical challenges of the TAPS design and implementation and the experiments with users conducting tactical field operations supported by TAPS. Most recently, we applied TAPS in a field experiment to support tactical user collaboration on pop-up tasks, sharing situational awareness, orchestrating tactical assets and services, and handling service failures that occur frequently in fast-tempo tactical situations characterized by quickly changing networking conditions and continuous movement of users. We present and examine TAPS from the aspects of scalability, performance, network usage, and under several different use case scenarios. Finally, we discuss the current limitations, and our lessons learned from applying multi-agent pervasive service collaboration technologies for real-time tactical pervasive service- oriented platforms. evolving tactical situations. Everyday, our technical approach faces challenges in handling situational awareness of both hostile and collaborative friendly forces combined with the factors of unstable tactical computing and network environments. To support such tactical operations, we identified that being adaptive in enabling tactical users to have faster and better decision making to deal with the ever-changing tactical situations is one of the most important capabilities for TAPS to provide. Being adaptive is not enough for TAPS to be useful to support the needs of real- time tactical operations. TAPS needs to provide real-time tactical users with the agility of delivering right information or services with context-awareness at the right timing. Though our current TAPS implementation is far from providing the full spectrum of real-time tactical operational needs, we have incrementally learned from our previous short experiments—our current field experiments shed light on how we should evolve TAPS. We now provide more technical capabilities while overcoming the issues imposed by hardware, software, networks, environments, and user interactions.