Affecto Oyj : Short AI enterprise stories

I wrote a blog about maturity of artificial intelligence (AI) and the critics provided by the World's most respected scientist a while ago (http://www.affecto.com/insights/blog/artificial-intelligence-not-yet-say/) Despite the academic opinion, AI, or machine learning is clearing its way to enterprises. This is a story of a few advanced applications of machine learning in use today - and the tips the organizations wanted to share in World AI Summit in Amsterdam in October this year.

Swedbank - Automated Customer Service

Swedbank has a clear vision - all client conversations start with a bot. The objective is to gather the intent of a customer in a very first step of a journey. Bot is designed to be a proactive system in order to give customer a choice, such as call customer back should one not receive the answer to the question.

The customer service receives approximately 45 000 requests monthly. The team of 35 people answered those questions. The most requests are about transactions related to cards, cash or payments to abroad, i.e. simple tasks, where the bot easily provides the information. Today, with the help of chatbot, three people are needed to monitor and manage the content - while AI will give 85 % correct answer. The most important learning Swedbank wanted to share from their experiences was:

• Invest AI only if it will save your money or make you money!

Booking.com - The Best Offer

The objective of booking.com is to take a hassle out of your travel! Their mission is to be 'customer first AI company'. It is today 7 largest internet company in the world, providing the best option for a traveller based on their needs.

Information retrieval is not easy and people interact with booking quite a long time. They deal with 1,4 million reservations per day and apply machine learning (ML) in several areas such as optimal bidding, destination suggestions, accommodation recommendations, fraud detection, demand prediction with 46 languages, and urgent topic detections - to name the most important elements of their services.

As far as we consumers are concerned, we don't even realise we are dealing with AI, it is not visible to us even if we are accommodated to use it. The lessons learned in Booking.com they want to share is;

• it takes more than ML skills to have impact
• the first hypothesis what customer needs is rarely correct - do rigorous testing to get there!

Vodaphone - Disrupt or be disrupted

Vodaphone's mission was to re-create their procurement via digital revolution, since they anticipated that in any moment in time a competitor will do some disruptive moves anyhow. The ultimate objective is to allow category manager to spend less time to analyse data and spend more time on decision making with the help of machine learning predicting demand.

They built a Digital Supply Chain Management House, to change the way people are working in their organisation. Company also realised that this is the way to attract smart people to work for them. Employee experience became a foundation block at the heart of everything they do.

The key learnings from their own disruptive actions were, that

• there is lots of bias in machine learning, so one need to be critical with our human values do not get misunderstood
• AI is truly multi discipline team work.

Phrasee - AI writes better than human

Having studied millions of words, phrases and sentences, Phrasee has proven there is a science to human language. Phrasee is designed to use artificial intelligence (AI) to generate email marketing language that outperforms humans!

They are using a deep learning model behind to detect which subject line in e-mail marketing works best. They claim human put their own believes and bias in these decision, yet AI can help to predict the one most effective one in marketing language. There is real science and data behind their decisions, not human intuition. This is proven by several their customers, such as eBay. Phrasee claims humanizing is not necessary, since human need to know it speaks to machine as it is in Swedbank's case. The learning of Phrasee's experience is that

• AI becomes powerful when it is invisible - that is when you know something good happened.

Netflix - Personalised offerings

Most of us are familiar with Netflix offerings, yet what we do not know, is how AI is working in the behind of scenes for the benefit of precision marketing. Netflix uses AI naturally for recommendations and personalized search, top rankings, trending now, continue watching - services, yet the more interesting use case of AI is for inspiring and engaging consumers. One example of such functionality is to profile customer based on historical behaviour and offer videos based on these profiles.

They even select the movie cover for you based on the profiling. Should you be identified as a drama enthusiast, you might get a recommendation to watch 'Top Gun' with the cover of Tom Cruise and Kelly McGillis in a cover picture in some romantic scene, while action movie lovers get the same movie promoted with pictures of pilots is a plane! This is called priming in marketing terms and we all fall for it!

eBay - Trust and personalization

eBay is a benchmarking company in many ways; it was the first global market place as well as first mobile on-line store. Today, more than half of eBay's business comes from mobile commerce. In eBay's on-line store, AI is about building the relationship with human. Their objective is to build trust with consumers. Algorithms learn from their customer's behaviour the preferences and intentions which leads to more personal interaction. While AI learns from consumers behaviour, it adds value by offering curated collection.

eBay uses AI to understand conversational, visual, emotive and personal modes of consumer, enhance company's understanding of inventory in real-world context and help to solve large information problems vital to understanding market and future commerce.

The academics are still critical on AI and machine learning, which is also part of their job to provide critics as scientist. Even though we deal with AI as consumers in several applications today. In top of mentioned above, we use AI fluently in Facebook, Google, Apple, and on-line stores, to name few. There is clear evidence that Business to Consumer (B2C) sector is well ahead in using algorithms in order to provide better customer experience!