An organization’s most valuable knowledge—its essential intellectual capital—is not limited to the information in ofcial document repositories and databases, such as scientic formulae, ‘‘hard’’ research data, computer codes, codied procedures, nancial gures, customer records, and the like. It also includes the largely undocumented ideas, insights, and know-how of its members (see, for example, Nonaka and Takeuchi, 1995; Stewart, 1997; Davenport and Prusak, 1997; Senge et al., 1999). This informal (often tacit) knowledge is deeply rooted in individuals’ experiences and the culture of their work communities. It commonly originates as practical solutions— through everyday inventions and discoveries—to the problems they must solve and thus serves as the critical resource for ordinary work practice (see, especially, Brown and Duguid, 1991, 2000). Much of this knowledge often remains embedded in practice. Small circles of colleagues and work groups commonly share crucial steps in a new practice and fresh solutions to recalcitrant problems through conversations and stories, with members lling in the background and gaps from their own experience. These groups and communities use the local vernacular to express these instructions and stories. Organizations face the challenge of somehow converting this valuable but mainly local knowledge into forms that other members of the organization can understand and, perhaps most important, act on. Here we present a detailed account of one organization’s effort to encourage inventiveness, capture new ideas, and use technology to then share the best of this knowledge beyond a local work group. Our account is based on our experiences during seven years with the design, development, deployment, and evaluation of the Eureka system at Xerox Corporation. Xerox uses Eureka to support the customer service engineers (CSEs) who repair the copiers and printers installed at customer sites. In four iterations, the system went from an experiment that researchers at the Xerox Palo Alto Research Center (PARC) designed to measure the value of codied eld experience to a system deployed to 20,000 CSEs worldwide. By focusing on communities and how they share knowledge in ordinary practice, we developed a set of questions and a methodology that we hope will enable others to build similar community knowledge-sharing systems. However, deploying any knowledge system involves pushing changes within a corporate culture; understanding the Eureka experience and the problems facing all knowledge systems to be deployed in the real world requires equal focus on these challenges. Our narrative covers the history of this project, carefully detailing the fundamental interrelationships between the social and the technical.We include a framework for building these kinds of community systems (see the sidebar) and our reections on the barriers to organizational change that their proponents confront.