Agents are programs or entities that operate with little or no human
supervision. They initiate actions, construct plans, migrate to
different locations, and communicate with other agents. Most importantly
they can independently respond to events and adjust their behavior
accordingly to accomplish goals. Well designed agents will have
personality (and like a good waiter will intrude only when necessary)
and remember training and tasks even if the user's computer crashes
or is turned off.
There
are different classes of agents depending on the agent's abilities:
they may be static or mobile; react to events or not; work alone
or with other agents; learn or be hardwired; autonomous or not.
Intelligent
agents can solve several classes of problems. They simplify distributed
computing, information retrieval, sorting and classification of
data, and handle repetitious tasks for users. Agents have taken
over many tasks people do not wish to do themselves, like scheduling
appointments, answering email, sorting news group information, and
getting the current news stories that match a person's interest.
As the agent learns more about its user it will become more useful.
Agents'
behavior and ability to solve problems may be either in the individual
agent or the agent may serve as a dumb part of a group that can
solve a problem. Agents that work as a part of a group form a more
stable system and may be able to handle tasks not easily done by
computers. Without a central intelligence the group may actually
become stronger and smarter. This type of agent setup may scale
up better than individual systems. Java and Python are the preferred
languages for agents.
White
papers on our products and technologies are available upon request.
E-mail us at innovation@bluetronix.net
or call 440.247.3434.
|