Redesigning Software Procurement through Intelligent Agents

Software procurement represents a process of vital importance to most enterprises in the public and private sectors, but particularly in the U.S. government, the software procurement process is highly pathological. Intelligent agent technology offers the potential to remedy this process pathology and an agent-based redesign transformation is recommended to dramatically improve the performance of the software procurement process. This paper outlines a number of contemporary agent applications and presents the results of a measurement-driven redesign analysis of the government software procurement process. We discuss intelligent software procurement agents that are developed to automate and support this process using a novel tool called the Agent Development Environment (ADE). The paper highlights the behavior and utility of intelligent software procurement agents and presents some preliminary results associated with their application to an operational government organization. We show how this agent-based process redesign offers excellent potential for improvement in both cost and cycle time for the process and offer an agenda for continued research along these lines. Redesigning Software Procurement With the passage of each product lifecycle, software becomes increasingly important o efficiency and effectiveness in the public and private sectors alike. Many major corporations rely on the power of information technology (IT) to compete effectively in this fast-paced, hypercompetitive, global business environment of today, and the military has 10ng invested in software-intensive weapon and communication systems. With an expanding mission but declining budget, the U.S. government is fast becoming software-dependent as well, as it represents the largest single buyer in the world with an estimated $40B worth of annual software procurement (STSC 1996). However, U.S. govemment procurement lead times are notoriously long and the time required for software purchases often exceeds the product lifecycles themselves. It widely employs linear, bureaucratic, paper-based, regulationladen procurement processes, even when buying the same commercial off-the-shelf (COTS) sottware used consumers, for example. Although the govemment has made considerable progress in terms of procurement over the last few years (e.g., acquisition streamlining, using EDI, eleclronic bulletin boards, emphasizing COTS software, establishing a preference for commercial specifications and standards), its procurement process will always be disadvantaged with respect to those found in the corporate sector. The immense size of the government organimtion, requirements for public Irust and accountability, socioeconomic legislation and other attributes all contribute to this result. In fact, the government must develop superior processes just to achieve parity with the private sector in terms of process efficiency and effectiveness. In this paper, we employ measurement-driven inference to diagnose a number of pathologies associated with the govemment procurement process and use the results to redesign software procurement through an intelligent agent application. This agent technology serves to enforce process integration between buyer and seller, takes advantage of global connectivity, satisfies regulatory requirements with even greater consistency than is possible today, and automates most of the software procurement process with attendant savings in terms of cost and cycle time. When implemented to augment the govemmenfs suite of legacy, current and emerging IT tools, this agent application can effect the kinds of quanUma performance improvements ought through process redesign (Davenport 1993). We provide a brief overview of some exemplary intelligent agent applications in the following section and then outline the key steps and results associated with redesign of the software procurement process. The architecture for our intelligent agent application is discussed subsequently, followed by the case of its initial application in From: AAAI Technical Report WS-98-13. Compilation copyright © 1998, AAAI (www.aaai.org). All rights reserved. the govemment acquisition domain. We close the paper with an agenda for continued research along these lines. Intelligent Agent Applications Work in the area of software agents has been ongoing for some time and it addresses a broad array of applications. Indeed, one need not research too far back in the literature to identify a plethora of agent exarnpleg zo many that any attempt to review them, even briefly, would constitute a journal-length paper in and of itsel£ In this section we provide a high-level overview of extant agent applications, with particular emphasis on a framework to relate them with this present work. It is informative to group extant agent applications into four classes: 1) information filtering agents, 2) information reWieval agents, 3) advisory agents, and 4) performative agents. Briefly, most information filtering agents are focused on tasks such as filtering user-input preferences for e-mail (e.g., Maes 1994, Malone t al. 1987), network news groups (Sycara and Zeng 1996), frequently asked questions (Whitehead 1994) arb. iwary text (verity 1997). Information relrieval agents address problems associated with collecting information pertaining to commodities uch as compact disks (Krulwich n.d.) and computer equipment (uVision 1997), in addition services such as advertising (PriceWatch 1997) and insurance (Insurance 1997). We also include the ubiquitous Web indexing robots in this class (see Etzioni and Weld 1995) along with Web-based agents for report writing (Amulet 1997), publishing (InterAp 1995) and assisted browsing (Burke al. 1997). Agents for technical information delivery (Bradshaw al. 1997) and information gathering (Knobloch and Ambite 1997) are not Web-based per se, but they perform a similar function. A third class of agents is oriented toward providing intelligent advice. Examples include recommendations for CDs (Maes 1997), an electronic concierge (Etzioni and Weld 1995), an agent ’qaost" for college campus visits (Zeng and Sycara 1995) and planning support for manufacturing systems (Maturana and Norrie 1997). Agents for strategic planning support (Pinson et al. 1997), software project coordination (Johar 1997) and computer interface assistance (Ball et al. 1997) are also grouped in this class, along with support for military reconnaissance (Bui et al. n.d.) and financial portfolio management (Sycam et al. 1996). Performative agents in the fourth class are generally oriented toward functions such as business transactions and work performance. Examples include a marketplace for agent-to-agent ransactions (Chavez and Maes n.d.) and an agent system for negotiation (Bui n.d.), in addition to the performance of knowledge work such as automated scheduling (Sen 1997, Walsh et al. 1997), cooperative learning (Boy 1997) and automated digital services (Mullen and Wellman 1996). The intelligent software procurement agents developed through this present research are probably best categorized in the fourth group above (i.e., performative agents), but they have been designed to also exhibit behaviors such as information filtering and reWieval, and their use can be accomplished through simulation (i.e., in an advisory role) as well as enactment (i.e., the performative role). Thus, intelligent software procurement agents have similarities with examples from each of the four classes above. To further descnl~e and differentiate intelligent supply chain agents, we have integrated the agent-taxonomy work of Franklin and Graesser (1996) with a three-dimensional structure from Gilbert et al. (1995) develop the analytical framework presented in Figure 1. Collaboration Parallel processing