Getting Started with Artificial Intelligence

"If you don't have an AI strategy you are going to die in the world that's coming." - Devin Wenig, CEO of eBay

How does my organization get started with Applied AI?

From our analysis of five sectors, and more than 50 cases, the team behind the Applied Artificial Intelligence-project have identified a number of overarching issues such as the challenge to implement the right AI technology and choosing the right platform.

On this page we have provided guidance on some subjects that require careful consideration, when an organization is to get started with AI.

AI blomst.png

1.pngTo answer this, you need an understanding of the available technologies, third-party solutions, and competitive pressures. If, for example, you are in the banking or utility business, you will have to consider applied AI in various aspects of the business; commercial, operational, central staff function, etc.

If there is a focus on commercial aspects, the opportunity could be on developing new models for churn reduction, on improving of acquisition or channel effectiveness. Furthermore, applied AI could support pricing decisions.

In case of an operational focus, contact center productivity could be improved via bot technologies, such as GupShup, or AI could, for example, be used to support decisions regarding retail location.

In the case of Bidgely, AI is applied to reduce churn and establish new customer engagement for utilities, which could potentially lead to more sales. "Disaggregation" is the name of the applied AI method here, which works by recognizing different patterns in electricity usage. Potentially, the technology could be used to identify, for example, the electricity use-pattern of a household fridge and tell whether the fridge is energy efficient. 

If the focus is on central staff function activities, solutions like AppZen for expense management could be considered.



2.pngOrganizing AI in an applied context requires new ways of thinking. Though optimal organizing depends on the specific case, our research generally shows a need for new organizational forms and smaller cross-functional teams.

In the case of Nordea, an independent, agile unit comprising of one subject matter expert, data scientists, and data engineers was established. The leader chosen for the unit, a very experienced manager, was found in-house, while the rest of the team was hired from the outside.

3.pngImplementation of AI is in most sectors still in the very early phases and an organizational mode of experimentation and piloting is key to get started. Our Applied AI examples vary in complexity from the implementation of third party solutions such as AppZen and Bidgely to developing in-house solutions based on Enterprise systems such as the credit security platform for Nordea.

How Pil Hvid.png

That said, implementing third party solutions require careful planning and preparation in most organizations. Bidgely, for example, needed not only to get access to smart-meter data, but also required a several month period to calibrate the AI model to the specifics of the utility customer base. In the more complex projects based on developing enterprise solutions, Ayasdi (featured in our finance section) encourages everyone to consider the key five pillars of discovering, predicting, justifying, acting and learning.  All elements are important and, in some industries, some are particularly challenging; i.e. the need for justification is key in the finance industry in areas such as AAI in credit application.

4.pngThe decision about the right technology is subject to the specifics of the organization and maturity in terms of applying AI. To get an understanding of the realm of opportunities, we encourage scouting and finding inspiration in other companies. One way to do it is, is following in the footsteps of what SE did when they visited Silicon Valley.

Look at relevant solutions offered by start-ups, investigate where the venture capitalists invest their money, and scout for ideas by researching and meeting with Silicon Valley-based peers or study the large tech companies leading the way in artificial intelligence.

When you have identified the relevant space and possible solutions, you must consider whether to build the platform yourself, like Nordea, or buy a third party solution.

5.pngData is always a key question in any project focusing on AI. Ayasdi emphasizes the focus on data availability over data quality and not letting your organization wait to get started but rather experiment in shorter cycles. In our Nordea case, the emphasis is also on the importance of understanding the regulation and the limits it sets.



How to build for Applied Artificial Intelligence with Danny Lange

Join us

If you are interested in obtaining more knowledge about AI, consider joining us on one of our Masterclasses or participate in the AI Academy.