In the age of digital business transformation, enterprises seek to increase their agility and speed of IT delivery. To accomplish this, they change their existing control-driven IT organizational structures and processes and establish separate modes for business-oriented and traditional IT delivery ("bimodal IT"). Though the concept of bimodal IT has been discussed in practice, empirical research regarding the approaches employed to implement bimodal IT is scarce. This paper presents findings from a qualitative-empirical study on the bimodal IT implementation approaches of nine companies. It identifies five different types of bimodal IT in these enterprises and shows that specific mechanisms are applied to enhance the (business) IT alignment in the respective organizational settings of each type. On the basis of similarities and differences among the types, we develop propositions for future research on bimodal IT and derive implications for practice.
During the last years, many companies established fast IT or digital IT units dedicated to build and operate digital customer-facing services. These units adopted agile methods, new tool chains as well as new organizational settings like BizDevOps teams. BizDevOps teams are responsible for continuously (re-)defining business functionality of certain (micro-)services, (re-)developing and running them. In these new fast IT environments, the role of enterprise architecture management changes dramatically. BizDevOps teams have a high degree of autonomy in designing both, the functionality and the architecture of their (micro-)services and thus contribute to business-IT-alignment in a new way. Nevertheless, a central enterprise architecture management (EAM) function is still required for supporting the teams regarding cross-team and cross-service issues. Furthermore, as many companies still run the traditional IT function side-by-side with the new IT function, the EAM functions of both parts have to cooperate. Based on a single case study, we discuss the emergence of a new EAM function ("fast IT EAM"), changing tasks and processes, implications for EA models and challenges for the integration of the traditional EAM and the fast IT EAM functions.
Today, Enterprises act in an increasingly interconnected world and in different kinds of collaborative networks. They are part of business ecosystems in which they interact with their customers, partners and competitors. The processes of analyzing and planning the intertwinement of business and IT architecture within enterprises has been successfully supported by enterprise architecture management (EAM) approaches. In this paper, we analyze four cases from different industries (health care, logistics, retail, and education) and argue that the intra-organizational concepts of enterprise architectures (EA) and EAM need to be extended to grasp the challenges of the enterprises' interconnectedness. Beyond the known concepts of extended enterprise architecture and federated architectures, we define five stages of extended architectures. Additionally, we describe challenges and existing solutions, which are relevant for this extended perspective.
Automation of incoming invoices processing promises to yield vast efficiency improvements in accounting. Until a universal adoption of fully electronic invoice exchange formats has been achieved, machine learning can help bridge the adoption gaps in electronic invoicing by extracting structured information from unstructured invoice formats. Machine learning especially helps the processing of invoices of suppliers who only send invoices infrequently, as the models are able to capture the semantic and visual cues of invoices and generalize them to previously unknown invoice layouts. Since the population of invoices in many companies is skewed toward a few frequent suppliers and their layouts, this research examines the effects of training data taken from such populations on the predictive quality of different machine-learning approaches for the extraction of information from invoices. Comparing the different approaches, we find that they are affected to varying degrees by skewed layout populations: The accuracy gap between in-sample and out-of-sample layouts is much higher in the Chargrid and random forest models than in the LayoutLM transformer model, which also exhibits the best overall predictive quality. To arrive at this finding, we designed and implemented a research pipeline that pays special attention to the distribution of layouts in the splitting of data and the evaluation of the models.