Summary: Business doesn’t want AI. Business wants results. While we were focused inward on our Advanced Analytic Platforms, smart competitors were rolling up AI/ML with other capabilities into “Intelligent Automation” platforms. This large scale integration of capabilities of which AI/ML is only a part looks a lot like the development of ERPs in the late 90s.
If you’re part of the AI/ML community it’s easy to get wrapped up in how AI/ML can have these effects for business but also easy to miss the fact that there are other tools that can also do this.
We may be getting the most hype but almost every article comes with a caveat for the business user about difficulties in implementation and shortage of skilled resources. So it’s important to remember that every business investment comes with careful consideration of risk.
While we’ve been focused on advanced analytic platforms, automated machine learning, and deep learning there are a variety of other ‘platforms’ based on preexisting capabilities that have been rapidly encroaching on our turf. It’s completely plausible that AI/ML may be swallowed up as a tool within these more business friendly packages.
The most current name for this alternate platform is “Intelligent Automation” or in some cases “Intelligent Business Process Management”. If you look at the Gartner or Forrester magic quadrants what you will not see with very few exceptions are the familiar names of any of the AI/ML platform leaders. Only IBM and Tibco appear in some but not all of these reviews.
What you will see if you look closely is an increased reliance on AI/ML capabilities to bring predictive analytics and text, voice, and image capabilities to these suites. Here’s roughly how this evolved.
Electronic Content Management Systems (ECM)
In the 90s and the early 00s Electronic Content Management Systems (aka Document Management Systems) were the most technologically advanced methods for in-taking paper like invoices and time sheets, converting them to structured data in electronic media, and automatically routing them for approvals and other types of processing. Even before DNNs these systems did quite a credible job of OCR (optical character recognition). The automatic and predetermined routing for action was the earliest precursor of RPA (Robotic Process Automation). These systems were extremely widely adopted and it’s no surprise that several of these major players show as key competitors in modern Intelligent Automation platforms.
Business Process Management (BPM)
Growing out of the Process Improvement and Process Value Analysis movements of the 90s, BPM modules began to appear as both standalone software as well as integrated with major ERPs like SAP, PeopleSoft, and Oracle. The no-code UI-friendly modules allowed teams at the LOB level to define, improve, standardize, and monitor business processes. These typically did not include any elements of Robotic Process Automation.
Robotic Process Automation (RPA)
RPA is today’s hot ticket for beginning one’s digital journey. It’s a mature, inexpensive technology that’s relatively easy to implement and maintain.
RPA is no-code UI-friendly standalone or integrated software. This is simple rules-based if-then-else software meant to automate the simplest repetitive processes to save on manpower and increase speed and standardization. It does not include BPM capability which continues as a parallel and still valuable application.
Since RPA is commonly used in call centers it frequently incorporates chatbot functionality. Fortunately or unfortunately this has caused surveys by McKinsey and others to include RPA in scoring whether or not a company is using AI/ML. To my way of thinking, the inclusion of a chatbot alone in this type off app does not prove you are thinking strategically about AI/ML and dramatically inflates statistics about companies who are strategically incorporating advanced analytics.
A few of the standalone RPA suites are beginning to include some predictive capability for example to embed next best offer in their customer responses. The leaders in standalone RPA suites do not include any of the leaders we would recognize from advanced AI/ML platforms.
Intelligent Automation / Intelligent Business Process Management
There’s a bit of a naming war going on among rating agencies but both these titles mean the same thing. Most importantly they bring together in a single application all of the preceding apps plus, increasingly, AI/ML capabilities.
This graphic from the Everest group shows very clearly how the many components of enterprise automation are being brought together in these single IA platforms. Note specifically the inclusion of AI/ML.
And this additional graphic from Everest shows how all the components of the earlier applications plus new synergies are brought to bear.
A Lesson from History
This rollup of increasingly sophisticated and integrated capabilities is exactly like the period leading up to today’s modern ERPs from SAP, PeopleSoft, Oracle, and others.
It began with the application of computer automation to targeted business functions using custom code. It progressed through Reengineering (custom code writ large and expensive) to the earliest standardized packages. These early cross-industry packages for MRP, Finance, and HR were the first to incorporate best practices and allow the enterprise to largely off-load the maintenance and improvement of the software to large, reliable, independent developers.
It was only a short step from there to integrate these different packages into the massive ERPs that came to dominate.
It was probably less than 8 years from the beginning of this process to the solid emergence of the fully integrated ERPs. I suggest we are seeing something like this today in data science. While we’ve been focused narrowly on our fun backyard, several smart platform developers are bringing all of our technologies plus tech like BPM and RPA into full integration. Business likes that.
Here’s the Gartner Magic Quadrant for this group:
Aside from IBM and Tibco, where are the AI/ML leaders? This may be like the stealth strategy of Platformication. If you weren’t looking carefully at how your customers were behaving, you could be quickly swept away.
About the author: Bill is Contributing Editor for Data Science Central. Bill is also President & Chief Data Scientist at Data-Magnum and has practiced as a data scientist since 2001. His articles have been read more than 2 million times.
He can be reached at: