Cloud computing systems supply massive infrastructures for high-performance computing that are flexible as they are able to acclimate to user and application requirements. Cloud Computing offers on demand services which are used by virtue of a service-oriented interface that execute the anything-as-a-service archetype. The enlargement of Cloud computing has resulted in the establishment of large-scale data centers around the world containing thousands of compute nodes and these data centers consume excessive amounts of electrical energy resulting in high operating costs. Therefore, to cut down the cost of energy consumption the cloud providers must optimize resource usage by performing dynamic consolidation of virtual machines (VMs) in an effective way to improve energy efficiency in cloud data center. The problem of VM consolidation can be split into four sub-problems namely physical machine overload detection; physical machine under-load detection; VM selection and VM placement. Each of the afore-stated subproblems must operate in an optimized manner to maintain the tradeoff between energy and performance. In this research paper a new multi-agent system (MAS) for dynamic consolidation of VMs is proposed with the aim of making the cloud system smarter by incorporating the five traits of multiagent systems which are ubiquity, interconnection, intelligence, delegation and human orientation. The Cloud Computing systems require intelligent and perceptive based software with progressive, elastic, self-ruling style which can be provided by MASs. The proposed method has significantly reduced energy consumption and at the same time ensures a high level of constancy to the Service Level Agreements (SLA).