This paper discusses the potential for personalized, user-owned decision-support systems. It can be readily seen that there are benefits from analysis of ‘Big Data’ that could not be attained through more traditional means, e.g. insurance and credit card fraud can be detected more readily when it is possible to analyze integrated data across multiple servers owned and controlled by separate organizations. However, high-level data analysis, though useful, cannot be trusted to provide all the answers to organizational ‘questions’. Individuals need to be able to inform themselves in complex decision situations and for this purpose there can be no substitute for ‘little data’ from wherever this is to be drawn. We explore a potential type of support that could overcome the barriers to professional creativity arising through lack of trust in decision-support systems owned and controlled from senior management. The Virtual Personal Assistant described uses natural language processing to interact with a professional user in the context of messy, situated problems, and in private. It has capability to learn from user-interactions and therefore to co-evolve contextually. A ‘little data’ system such as this can therefore help to improve relevance of user understandings in a relatively risk free environment.