Our Articles

A Monthly Article from our Speakers

Current Article of the month

Articles of 2017

Articles of 2016

Articles of 2015

Articles of 2014

Articles of 2013

Articles of 2012

Articles of 2011

Articles of 2010

Articles of 2009

Articles of 2008

Articles of 2007

Articles of 2006

Articles of 2005

Articles of 2004

Articles of 2003

Articles of 2002

Barry DevlinAll aboard the Digital Business! Oncoming IoT and AI

by Barry Devlin

July 2017


If you thought you were already on a fast train from BI to analytics and big data, I have news for you. Coming down the track at 300+ km/hour is a Frecciarossa of algorithms, Internet of Things (IoT), and cognitive decision making. You need to get aboard. And quickly. The challenge is that, in the tradition of James Bond, you must leap from one moving train to the other.

It is a mere five years since Apache Hadoop released its 1.0.0 version and opened the floodgates of big data. Hardy pioneers had blazed the trail a few years earlier, but the idea that maybe—just maybe—relational databases were not the only way of supporting business decision-making needs really only started to bear fruit in 2011. Since then, version 2.2.0 (October 2013) began to convert Hadoop into a “data operating system”, while September 2016 saw the 3.0.0 alpha release .

Why should you care about this history? Because, it is becoming history. Driven by then unimaginable quantities of social media and web commerce data, the Hadoop software ecosystem on commodity hardware offered the only affordable platform to store and explore this data. You could call this analytics in the traditional mode, but bigger. In fact, it was only in version 2.2 that batch processing could be superseded by online or real-time data processing. Big data mining is fine in batch, but applying the models in production demands modern processing with much lower latency.

In the past year or so, the ominous rumblings of various analysts have been heard: “Hadoop is dead. Long live…” Spark, Storm, Kafka and others have been anointed as the new kings. Iterative algorithms and interactive, exploratory analytics reign. The mantle of power moves from data at rest to data in motion. Overwhelmed by the speed and size of data from the burgeoning IoT, there may not be time to land it on disk or even to move it from the edge to the central environment before deciding its value or exploring its meaning. The traditional data warehouse architecture is no longer sufficient. We need a multi-pillar, multi-technology, fully distributed but deeply integrated platform, as described in my 2013 book, Business unIntelligence.

All the above, however, occurs only in the economy, IT carriage of the Frecciarossa.

It’s within the business class carriages that the truly revolutionary thinking sits. Gartner christened this digital business as long ago as 2014, defining it as “the creation of new business designs by blurring the digital and physical worlds” . It’s a misleading name, given that it encompasses only one half of the idea, and emphasises bits rather than business designs. The foundation is back in the IT carriage: big data analytics and the IoT stand for the digital and physical worlds respectively. How corporate strategies and behaviour must and, indeed, will change is the real news from business class.

Business, in one sense, is a very simple process: understand people’s needs, decide how to satisfy them and act accordingly. These three steps—understand, decide, and act (UDA)—form a tightly interconnected and closed loop in the digital business. Information/data from the physical world of business processes, social media and, increasingly, the IoT is the basic feedstuff of this cycle. Its outputs are the goods and services of the business. And its foundation is intelligence—the ability to reason, intuit and empathise about the relationship between input and output.

How business navigates this UDA cycle is about to change dramatically and irrevocably. We are crossing the threshold between decision-making support and actual decision-making/action-taking software. Artificial intelligence (AI), machine learning (ML) and cognitive computing are about to take “one giant step for mankind” and move that cycle from human wetwear to silicon software.

Simple, rule-based algorithms have been doing basic decision-making for some time now. Operational BI and decision management applications have been used for some years to automate suitable decisions. If the timescale of the decision is shorter than humanly feasible or its value less than that of human attention, and if the inputs and outputs are related by straightforward logic, algorithms are already widely used. Originally rule-based or table-driven, such algorithms have more recently been extended or replaced by more sophisticated, model-driven ones, that emerge from the statistical and related analytical work of data scientists. While undoubtedly valuable in many cases, the true direction and destination of this train will be seen in the next advance in AI technology: the application of advanced machine learning techniques to more complex business problems and decisions.

The evolution of these applications is still at an early stage, but as they begin to gain traction, uptake could be rapid and deployment widespread because of their benefits to the bottom line. We already see AI applied in fraud detection, portfolio management and underwriting in financial services. Major strides are occurring in healthcare: in diagnosis from imaging technologies, as well as disease diagnosis and prognosis. In manufacturing and utilities, forecasting and problem anticipation are being improved with ML. These examples illustrate the beginnings of a trend from automation of deterministic decisions to augmentation of human intelligence in ambiguous situations.

In terms of the traditional decision-support pyramid, AI and ML is becoming widespread in operational BI, is expanding into tactical BI, and showing initial promise in strategic BI. As we move up this pyramid, the role of AI becomes broader, more ambitious. The business opportunities increase too. Automation speeds response and reduces costs of employment. Augmentation identifies and enables new income and new markets.

Consider the discussion in the Frecciarossa business carriage. Omega S, a descendent of Alpha Go (Google DeepMind’s world champion beater ), discusses—yes, “she” listens and responds—business expansion opportunities with a CEO in 2020. Omega S has accessed and deeply analysed years’ worth of market sentiment and economic forecasts, the published results and outlooks of all competitors, as well as detailed internal data. She quickly draws likely financial and other outcomes of several scenarios on the screen. The CEO’s original option is way down the list of likely success. The AI-generated option is the obvious winner. How could he have missed it? Should the CEO accept the suggestion, challenge it, or simply resign?

The question for you is more immediate. How are you going to leap from your current train—now looking slow and shabby—to the Frecciarossa? It’s not going to stop and wait for you.

Upcoming events by this speaker