This paper tries to describe a coherent vision of a possible next generation of HIS that rethinks how and what for computing is used in hospitals. Current systems, organized mainly around the ‘database’ and the ‘communication’ paradigm help the data processing to a great extent. At the same time they cannot be accepted as ‘automated’ systems, as handling of information is done mostly by end- users, i e. by human actors. Emerging methods enabling automated information handling are the following: integrated handling of different media, seamless communication among different systems, alternative input-output devices, tools for pro-active information handling These technologies should be grouped to two main branches: technical advances in data handling and theoretical advances in knowledge handling. The advances in knowledge handling are really important: tools based on that can take over routine information handling tasks from human end users. To discuss automation in HIS it is useful to understand the process of information handling in general within the hospital. A suggested multidimensional information space, where information objects are gathered mainly along two axes, the ‘patient axis’ and the ‘management’ axis might be of help. Combinations of selected dimensions resulted in a space of an estimated 2294082 possible information handling situation types in an earlier publication. Automation of information handling tasks can be derived from this model. The authors suggest to automate certain tasks done usually by active actors of the information handling situation space. Software agents working ‘on their own’ are known entities in HIS systems. Two components are needed for automation:

– an organized data base where its content can be ‘understood’ and interpreted by an algorithm, with other words a knowledge base

– an algorithm, that covers a certain routine information handling task.

Data bases of HIS should be re-thought in a way that enables automated processing to a greater extent. The development of data base technologies clearly point to this direction. If most of the data bases of a HIS will be like that, new generation of applications might be launched to use them. E.g. a software agent called ‘patient assistant’ could collect data from different sources and build a coherent and updated patient file. The results of a knowledge based, agent operated HIS should be the following:

– significantly less direct human involvement

– significantly less paper to be produced

– enhanced speed of data flow in general

– enhanced reliability by widespread watchdog functions