The Emergence of Application-Specific IoT

By April 24, 2017Internet of Things
Delicate Tension (Wassily Kandisnky, 1923)

In a recent announcement, covered here under the title PTC’s ThingWorx and Deloitte Release Plans for Industry-Specific IoT Solutions, the two companies have announced a plan to collaborate on industry-focused IoT initiatives.

Back in September 2014, which, in IoT-years, seems like a decade ago, I published a blog article under a very similar title, anticipating The Emergence of Application-Specific IoT.

Below is an abbreviated and edited excerpt from of my 2014 article, followed by additional commentary. You will be the judge of how much, or, perhaps, how little has changed in nearly 3 years of intense activity in the IoT space.

The Emergence of Application-Specific IoT

In Gartner’s 2014 Hype Cycle Special Report, which evaluates the market perception and penetration for over 2,000 technologies, services and technology trends, of particular interest to many was the placement of the Internet of Things (IoT) at the top of the Peak of Inflated Expectations.

Indeed, inflated market size forecast and grandiose visions of IoT reign supreme. Morgan Stanley forecast the by that by 2020 there will be 75 billion connected devices, whereas Harbor Research estimates the same market will consists of a meager 8 billion connected devices.  Gartner estimates that by 2020 there will be 26 billion connected devices, IDC counters that with 30 billion connected devices, Cisco says 50 billion… you get the picture.

Do you think they really know?

On the other hand, does the number of connected devices matter that much?

What really matters is the data these devices can collect and the business value they provide when connected to each other and to advanced analytics and decision-support applications. , i.e. the ability to achieve a sustainable business value.

But it appears that many in the press and in industry do not subscribe to Gartner’s assessment and timeline. The IoT gets much attention, ranging from breathless headlines to long term strategy decisions by companies that span the gamut from onboard data acquisition and wireless communication hardware to enterprise software.

Regrettably, there are still as many trivial IoT scenarios as there are serious ones and, by and large, we are still awaiting the realization of the exponential growth in IoT revenues the pundits are promising us. Even investors that too often pursue early technologies and unsubstantiated business ideas, sometimes inexplicably, are lukewarm about the potential value of the dozens of companies in the IoT space, most of whom are generating less than $10M in annual revenue.

How will IoT fans, pundits, and especially company CEOs betting their company business on IoT explain the disparity between the vision and the reality? We are already witnessing the introduction of new “classes” of IoT technology such as the “industrial IoT.” I expect that we will witness the emergence of “application-specific IoT”, which, in many ways, will not be that different from the many Machine to Machine (M2M) technology implementations that address a narrow, yet no less valuable business opportunities such as the solutions implemented, for example, by Axeda (now a PTC product) at Diebold, EMC, GE Healthcare, and Philips Healthcare.

Why Application-Specific IoT?

Manufacturing and service companies are adopting outcome-driven business models and using the Internet of Things to redefine product offering and engagement models that go beyond operational efficiency improvement. They leverage IoT to implement profitable growth opportunities.

Leading companies will transition from a product-centric mindset to a collaborative and interconnected model in which success hinges on how well they can integrate their organization as part of a complex ecosystem and a dynamic network of producers and consumers of information and services that will align itself around outcomes and innovation.

The value-chain participants, both internal and external, are semi-independent subsystems, each with inputs, transformation processes and outputs. They employ vastly diverse business practices and operating models, acquire and consume different resources, and are subject to different, sometimes conflicting, business goals. And they operate across geographies and cultures and are subject to different regulations that influence the ability of disparate units to work in harmony.

An IoT platform is a configurable and dynamic value-driven framework that effective and efficient collaboration among value-chain partners and stakeholders to exchange value for the benefit of all participants in the context of a specific business solution.

The platform provides a means for multiple stakeholders to collaborate and communicate – before, during, and especially after a well-defined business process is deployed. It provides in-context governance rules and manages partners’ and users’ identities, data rights and privacy by ensuring provides process visibility and traceability across value chain functions and business processes.

Some Thoughts on Data Analytics

We are seeing a surge in data analytics software that promises unprecedented abilities in data patterns discovery, trend analysis and even prediction. No matter how powerful these methods might be, the analysis as of what’s important and what’s noise (from a business decision point of view), what’s a correlation but not a causation, must be done within a business environment and in an application-specific context. And these findings, correlations and recommendations must be considered in the broad context of a myriad business rules and regulations, other, non-IoT data, and just plain best practices.

What about machine learning? The idea of machine self-learning from relevant data is powerful and we will no doubt continue to see impressive progress. But, by definition, machine learning requires good, up to date, vetted data. See, for example, what happened once IBM’s Watson was given access to the Internet and added The Urban Dictionary to its vocabulary.

Most organizations have not spent the necessary time and resources to properly collect and validate the corpus of broad knowledge they possess, both formally and in the grey matter between employees’ ears. And, whatever knowledge they did capture tends to be organized by rigid disciplines, is hard to locate, and it often becomes stale very quickly.

Organizations that wish to leverage the progress in machine learning and similar information-centric artificial intelligence methods (in contacts to first-principles reasoning approaches) will have to ensure high-quality application-specific information is exploited during systems training and throughout the product lifecycle. And since diverse data provides learning systems a broader context to identify patterns and provide deeper insight (The urban Dictionary excluded), organizations should consider expanding the definition of the Internet of Things IoT beyond physical “things” to include enterprise data systems and human input via social media, and employing master data management (MDM) systems more aggressively.

Image: Delicate Tension (Wassily Kandisnky, 1923)