This project works in the field of research of self-performance techniques for multi-agent systems. A cognitive mechanism for individual agents is aimed to be created so that they can be provided with self-performance patterns and an intelligent environment. This way, they could be applied to real-life problems that may need constant adaptation. That is the reason why  the very aim of this project is to develop an intelligent system which adapts the work manager planner dynamically in a high performance computer center. This architecture should be available also to other centers. The aim of the planner is to satisfy the administration department of a supercomputer center, which, in turn, has to satisfy the users and a series of policies that govern in the center, based on the optimization of (both internal and externally contracted) resources. In order to adapt this planner, different strategies based on multi-agent architectures where every agent will have learning and evolution capacities will be studied and settled. As the interaction process moves forward, the agent is intended to model the performance of the user or  the asigned resource, as well as the level of satisfaction of the aforementioned user/resource against different possible strategies of the planner. The planner's intelligent control system will use this information to generate the right strategy at every moment by means of an evolutionary optimization system.

The primary aim is, in turn, composed by partial aims:

• The establishment of a system of direct and induced measures to satisfy both the users and the computer center itself.
• The development of an architecture based on intelligent agents that model the usage of resources in the supercomputer center.
• The development of an architecture based on intelligent agents which model the supercomputer center's own resources, including the ones given by the centers  not involved but working through agreements or contracts.
• Extension of MDB (Multilevel Darwinist Brain) for self-performance agents. Agents will thus be allowed to operate in services architectures, improving the efficiency and management simplicity in these highly complex systems.
•  The development of a series of interfaces among the system components so that every component can be set up separately.
• The application of intelligent computer techniques to a commercial planner.
• The efficient cooperation between n-cluster and CESGA.
• The increase of the level of satisfaction of CESGA, followed by an increase in the level of satisfaction of the users and the level of optimization in the usage of computer resources for the intelligent planner and the developed architecture of planners. 
• The establishment of the nuis system.