A multi-agent system for sales order processing

The increasingly distributed nature of sales order systems results in some nagging, pressing, and crucial issues. These can be solved by the use of agent-based architectures. Growth of the Internet also has opened up possibilities for improving production logistics. Embracing the Internet for order processing can lead to cost-effective, fast, and flexible operations. So the trend now is to treat each customer order separately and make the best use of the Internet to process the order. This is easier said than done. The behavior of elements in the supply chain can no longer be preplanned. Consequently, these elements should have the intelligence to interpret the order, the ability to perceive the resource environment, the autonomy to take rational decisions, and reactivity to change their behavior to adapt to the environment. This is where agent technology can help the manufacturing community. At Infosys work is going on to develop an agent-oriented framework for sales order processing. The framework, which we call Agent-Based Sales Order Processing System (AESOPS), allows logistics personnel to conceptualize , design, and build a production environment as a set of loosely coupled distributed units over a number of physical locations. These production units can interact with each other to process any order to completion in a flexible yet consistent and efficient manner. We will discuss a few salient features of the framework. Desirable Features In a typical sales order processing application a number of stages are involved. The process is initiated with a customer placing an order. The order is reviewed by the Logistics department (See Figure 1). If the product is in Introduction " Customer is king. " This is more evident now than ever before for all industries, manufacturing in particular. Traditionally, manufacturing companies have based their production facilities at a small number of locations. The demand of the product was forecast using sophisticated forecasting methods, and a production plan was prepared accordingly. As long as customers were willing to select from a predefined set of models manufactured by the company, things went smoothly. In today's global and more competitive markets, customers have started demanding customized products. This has led to a situation where erstwhile planning methods go haywire. inventory it is shipped; if not Logistics prepares a plan for raw material inventory and production units for different processing stages taking into account the capacity, resource and time constraints. Jobs are often grouped to reduce cost. …