In today’s world, databases and database systems have become an essential component of everyday life, so much so that a life without DBMSs has become inconceivable. This article focuses on relational database management systems in particular, and proposes a novel and innovative multi-agent system that would autonomously and rationally administer and maintain databases. The proposed multi-agent system tool, ADAM, is in the form of a self-administering wrapper around database systems, and it addresses and offers a solution to the problem of overburdened and expensive DBAs with the objective of making databases a cost-effective option for small/medium-sized organizations. An implementation of the agent-based system to proactively or reactively identify and resolve a small subset of DBA tasks is discussed, and the GAIA methodology is used to outline the detailed analysis and design of the same. Role models describing the responsibilities, permissions, activities, and protocols of the candidate agents, and interaction models representing the links between the roles, are explained. The Coordinated Intelligent Rational agent model is used to describe the agent architecture, and a brief description of the functionalities, responsibilities, and components of each agent in the ADAM multi-agent system is presented. Finally, a prototype system implementation using JADE 2.5 and Oracle 8.1.7 is presented as evidence of the feasibility of the proposed agent-based solution for the autonomous administration and maintenance of relational databases.