There is no point taking action just for the sake of it. We expect peak hype to be followed by the typical "trough of disillsionment" soon. Analyze the impact of AI and start test-driving AI in a modestly sized pilot that can scale well if successful. For experimental projects, failure should be an option. there are an estimated two to three years for this learning phase.

The value of the technology derives from its potential to distil meaningful intelligence from data, making predictions on the basis of this data and therefore optimize decisions. AI helps enterprises to reduce costs, offer their customers better service, or design products more closely around what customers want. In the mid- to long-term, focus your attention on strenghthening customer orientation through AI.

Enterprises should hone their ability to scan for new business model opportunities and drive their own innovations. AI also needs deep integration with internal processes. Companies thinking beyond use cases in isolation from other are poised to upend an idustry, create new product categories and shape new business models. However, it is vital to stay aligned with the core of the enterprise. AI initiatives can only deliver genuine value when aligned with the core of the enterprise - they must fully embrace and extend it.

The DIY approach to building competencies will only work for a very few enterprises - and most of these will have an affinity with the tech sector and will already have specialist expertise in-house. The start-up scene has brought a huge variety of solution packages onto the market which are an easier and much faster route for many enterprises than developing and building their own technologies. And commit to a clear AI investment strategy: By 2020, AI will rank among the top five priorities of CIOs.

Businesses need to create a future basis for capturing comprehensive and high-quality data from customers, processes and machines now. If you have already introduced AI on a widespread basis for new processes and business models, what matters now is encouraging people to take AI on board as integral to the organization's way of doing things. Start with trials and channel the lessons learned into adapting the core processes. By 2022, businesses should be ready and prepared to fully implement AI.