The Gemini Multi-conjugate adaptive optics System (GeMS) at the Gemini South telescope in Cerro Pachon is the first sodium Laser Guide Star (LGS) adaptive optics (AO) system with multiple guide stars. It uses five LGSs and two deformable mirrors (DMs) to measure and compensate for distortions induced by atmospheric turbulence. After its 2012 commissioning phase, it is now transitioning into regular operations. Although GeMS has unique scientific capabilities, it remains a challenging instrument to maintain, operate and upgrade. In this paper, we summarize the latest news and results. First, we describe the engineering work done this past year, mostly during our last instrument shutdown in 2013 austral winter, covering many subsystems: an erroneous reconjugation of the Laser guide star wavefront sensor, the correction of focus field distortion for the natural guide star wavefront sensor and engineering changes dealing with our laser and its beam transfer optics. We also describe our revamped software, developed to integrate the instrument into the Gemini operational model, and the new optimization procedures aiming to reduce GeMS time overheads. Significant software improvements were achieved on the acquisition of natural guide stars by our natural guide star wavefront sensor, on the automation of tip-tilt and higher-order loop optimization, and on the tomographic non-common path aberration compensation. We then go through the current operational scheme and present the plan for the next years. We offered 38 nights in our last semester. We review the current system efficiency in term of raw performance, completed programs and time overheads. We also present our current efforts to merge GeMS into the Gemini base facility project, where night operations are all reliably driven from our La Serena headquarter, without the need for any spotter. Finally we present the plan for the future upgrades, mostly dedicated toward improving the performance and reliability of the system. Our first upgrade called NGS2, a project lead by the Australian National University, based a focal plane camera will replace the current low throughput natural guide wavefront sensor. On a longer term, we are also planning the (re-)integration of our third deformable mirror, lost during the early phase of commissioning. Early plans to improve the reliability of our laser will be presented.