There are a lot of different definitions of the agent. Here, in MagentA, we use this one:

Agents are the software objects capable of competing and/or collaborating among themselves with a view to achieving given tasks. Agents operate within a Virtual World, which represents a model of a domain of the Universe. Knowledge about this domain is stored in Ontology.

Agents compose messages, send them to other agents or to their human clients, interpret received messages and make decisions how to act upon them.

In agent-based applications, the term Ontology refers to the description of an environment (business, physical, economic or any other) within which agents operate. Ontology contains objects upon which agents act and rules which they must follow.

For example, Ontology comprising corporate knowledge forms the basis of agent activities in the MagentA i-Enterprise System. It describes the structure of a company, roles of its employees, main directions of its business activities, goods produced and sold, current and potential clients of the company and many other important aspects of the business. This knowledge represents the intellectual capital of the corporation, which is constantly being modified, deleted, updated, verified and replaced, developing and evolving along with the corporation.

MagentA products use Ontology in the form of a semantic network to describe subject domains and specific applications. This network uses fundamental concepts such as object, process, property, attribute and relation, to structure the relevant knowledge. The use of such a network allows us to separate knowledge of a domain from the source code governing agent runtime behaviour and to re-use it for a range of applications.

Unlike most of the well-known systems MagentA offers quite a new approach and a selection of software tools for developing agents and their communities.

Firstly, MagentA Corporation offers agents that are not mobile in the traditional sense of the word, as Java-agents. The agents of MagentA Corporation inhabit the user's workstation, server or networks of workstations/servers; they are developed in highly efficient languages like C++ or Object Pascal. This allows us to considerably increase the number of agents that can work in parallel (several hundreds of thousands), increase the speed of communication between agents (up to 50,000 messages per second), and to simplify or eliminate a whole range of design problems, including the problems of security, agent interaction etc.

Secondly, the users are offered agents with a whole range of intelligent abilities - from simple agents at the elementary level and quick smart agents to quite complex ones that are truly intelligent. The complex, intelligent agents are built in a new way: by the networking of simpler agents rather than by an extensive expansion of capacities and capabilities of a basic agent, as is traditionally done.

MagentA agents are result of an extensive research and development work in the field of self-organising agents and agent communities, virtual and abstract worlds, strategies for reasoning and simulation of negotiations based on the method of compensations, ontology architectures and building of multi-tasking runtime software environments.

MagentA multi-agent systems are characterised by emergent intelligence, flexibility, effectiveness, short development lead times and low-cost ownership, which makes them accessible to a wide range of potential customers.

     Autonomous operation 24 hours a day;

     Pro-active marketing, advertisements and sales;

     One-to-one dialogue with users;

     Ability to work under conditions of uncertainty;

     Scalability based on parallel execution.