Recent research has shown that educational games positively motivate learning. However, there is a little evidence that they can trigger learning to a large extent if the game-play is supported by additional activities. We aim to support educational games development with an Agent-Based Technology (ABT) by using intelligent pedagogical agents that can intervene to offer hints, assistance and suggestions when the learner is lacking knowledge, but does not intervene otherwise, so as not to interrupt game flow. In this paper we describe the possibilities of using pedagogical agents to infer learner's motivation and emotional state as they allow communication and interaction in a digital learning environment. Our approach emphasizes on improving pedagogical agent interactivity: from pedagogical agent techniques to "Tutor and Tutee" Agents' techniques to create a high social and collaborative Digital Educational Game (DEG) environment. [For full proceedings, see ED557189.]