From sophisticated single agent in complex environments to multi-agent system (MAS) organizations, intelligent software agent research has come a long way in just under two decades. Many new branches of research in this field have emerged over the years which have enabled today's agents to perform a wide variety of human-like tasks such as learning, reasoning, negotiating, self-organizing and trusting each other, etc. Unfortunately, very few practical MASs have been deployed after such a long period of intensive research and development. For MASs to achieve higher popularity among end-users, we believe that agent oriented software engineering (AOSE) should adopt a new paradigm as has been done in Web 2.0 -to allow endusers to actively participate in developing or modifying features in agents at various stages of the agent's lifecycle. In this paper, we propose a vision for democratizing AOSE. We discuss what potential new researches need to be carried out in the areas of AOSE and agent learning in order to realize such a vision of moving from MASs to mass end-user agent development, and discuss potential challenges facing various aspects of this vision.