In 1993, Yoav Shoham's paper on agent-oriented programming was published in the Artificial Intelligence Journal. Shoham's ideas, and the work on agent-oriented programming it inspired, has had a profound impact on the field of multiagent systems, as evidenced by Shoham's paper receiving a 2011 IFAAMAS Influential Paper Award recognising seminal work in the field.

Agent-oriented programming offers a natural approach to the development of complex systems in dynamic environments, and technology to support the development of agents and multiagent systems is beginning to play a more important role in today's software development at an industrial level.

Since Shoham's initial work, a range of platforms that support agent orientation have become available, and considerable experience has been gained with these platforms. Some key issues have also emerged from this work, however. First, given the plethora of systems and approaches that have become available in the field for developing multiagent systems, it is no longer clear which of these technologies is most appropriate for developing a particular application or what the distinctive benefits of various approaches are. It is especially important for practitioners to understand the benefits resulting from a particular choice of technology, when and how to apply it, and to develop standards that support the application of agent technology. Secondly, the very different style of agent-oriented programming potentially hampers the uptake of agent development tools and methods. To successfully apply the agent-oriented paradigm and to support the implementation and testing phases of agent-oriented development it is therefore very important to establish best practices and evaluate lessons learned from applying the technology in practice.

The aim of this seminar was to bring together researchers from both academia and industry to identify the potential for and facilitate convergence towards standards for agent technology.

The seminar was very relevant for industrial research. The seminar meetings were meant to enable interaction, cross-fertilisation, and mutual feedback among researchers and practitioners from the different, but related areas, and provide the opportunity to discuss diverse views and research findings. The interaction in a Dagstuhl seminar was considered to be ideal for establishing common ground for defining standards, identifying best practices, and developing approaches to applying agent technology to the large scale, realistic scenarios found in industry. The aim of the discussions that were planned was therefore to establish a future research agenda, i.e. a road map, based on an evaluation of current state-of-the-art of agent-oriented programming languages, tools and techniques that are particularly important for large scale industrial applications.

The seminar took place August 19th--24th 2012, with 37 participants from 15 countries. The programme included presentations by the participants and group discussions. Presentations were about 30 minutes long, including questions. We specifically asked participants not to present current research (their next conference paper), but rather asked for what should be considered the next step in their research area.

Participants were encouraged to use their presentations to provide input for discussion about the road map. They should show their perspectives and discuss what they think should be on the research agenda, try to explain why, and what it is they think this community should be aiming for. The group discussions took place in the afternoon, after the coffee break until 6 pm. We put together four groups of 8-10 members, each headed by one discussion leader. The results of each working group were then presented to all participants before dinner. The seminar concluded with a general discussion on Friday morning and a wrap-up.