Use conversational interfaces to analyse business data. The ability to ask conversational questions (e.g. "what is driving this correlation?") offers potential convenience, speed and reduced Business Intelligence costs. Will increasingly become a tool layered in to other applications / use cases.

Empower teams with data and tools to run advanced analyses on the business and adjacencies.

Personalised data feeds, potentially structured on a team / functional / hierarchical level will help speed up processes and decision-making. Differing organisational cultures around information-sharing will be a key issue here.

Reducing data training set requirements is a significant move towards lowering the cost and challenge of data deployment for modelling purposes, however it does have risks.

Building and maintaining high quality data for advanced analytics

Combine source data from different sources into meaningful and valuable information

Take data from raw formats with data quality problems and develop in to a clean, ready to analyse and deploy format. This may be as simple as mismatching Excel columns or the aggregation of completely different data sources and types. Linking object IDs is key - hence human reinforcement and tagging potentially part of the data management process.

Unless using unsupervised learning systems, high quality labeled data is critical. Labelling data minimises the risks inherent in using it. This can potentially be part-automated.

Generate cloned voice - at this stage largely for artistic, trouble-making and media purposes. This has potentially troubling implications for so-called "fake news" applications amongst many potential uses.

Avoid garbage in, garbage out by ensuring quality of data with appropriate data cleaning process

Predict commodity requirements - typically to optimise procurement strategy for large industrial organisations.

Optimise distribution network cost effectiveness (balancing capital and operating expenditure). Factors include route mapping, load balancing, capacity and demand analysis. Complicating issues may include weather and other exogenous factors.

A variety of AI techniques can be deployed to help capture insights on opposing sporting team approaches and strategies (for example recognise patterns in movement). AI can support modelling through alternatuive game strategies that coaches or managers wish to test.

Support data visualisation and use predictive analytics to help individuals and teams spot issues or opportunities for self-improvement - or weaknesses in opponents - to optimise athletic performance. Sensor data may be a significant element in the inputs.

Determine whether there are new business model opportunities to be examined- for example algorithmic deployment. These can be modelled and tested for potential scenarios.