Market research suggests that organisations, in general, have a differentiation strategy when approaching Electronic Commerce. Thus, in order to be useful, agent technology must take into account this market characteristic. When extending its application to the negotiation stage of the shopping experience, one should consider a multi-issue approach, from which both buyers and sellers can profit. We here present SMACE, a layered platform for agent-mediated Electronic Commerce, supporting multilateral and multi-issue negotiations. In this system, the negotiation infrastructure through which the software agents interact is independent from their negotiation strategies. Taking advantage of this concept, the system assists agent construction, allowing users to focus in the development of their negotiation strategies. In particular, and according to experiments here reported, we have implemented a type of agent that is capable of increasing the performance with its own experience, by adapting to the market conditions, using a specific kind of Reinforcement Learning technique.