The travel expense appraisal process is at the operational heart of any professional services firm, and relevant to many other large organisations. To put the importance of this process into perspective – alone, each of the Big 4 professional services firms employ more than 200,000 employees, most of which are travelling on a daily basis to provide their clients with a personal on-site advisory service and they’re submitting an overwhelming number of expenses every day. Any issues arising in this process will have detrimental effects on daily operations, and, because of the sheer volume of new expense reports being submitted daily, blockages in this process could very easily create an insurmountable backlog. In the example covered in this talk, this process was executed by 20 FTEs in the GSA region alone. This is where the value of AI can shine – through Deep Learning (DL), an AI can be trained to perform human-like activities, such as deciding the outcome of an expense claim from both structured data, e.g. the expense amount, and unstructured data, e.g. the description of and reason for an expense. The presented use case can thus be seen as an example of how DL can be used to automate data-centric processes, as long the AI system is tailored with care to the existing business process and expert logic. The two resulting key benefits are reduced operational costs (estimated at -70%) and an actual improvement in decisioning accuracy measured at ca. +25%.