Skip to main content

The top AI and automation trends to expect in 2018

(Image credit: Image Credit: Sergey Nivens / Shutterstock)

For years, artificial intelligence and process automation have evolved largely on independent paths. The coming year will bring steady innovation in both fields, but the real leap forward will be the synergy created when AI and intelligent process automation are harnessed together.

To explain what I mean, I’ll start by spending a few minutes on each area individually.

In AI and machine learning, high-value targets like speech recognition and synthesis, image recognition and bots were identified years ago, but have seemed to remain just out of reach. Now these capabilities are becoming widely appreciated due to advances in three critical areas: massive compute power, robust deep-learning algorithms, and the availability of huge volumes of data to train models.

Another adoption driver for AI and machine learning is that they are becoming more affordable and easier to consume as more vendors, including Microsoft, Google, Amazon, and others, offer these capabilities via the cloud. This gives almost any enterprise the ability to cost-effectively analyse vast amounts of data, be it related to its own processes, industry, or market.

We are also seeing progress in the development of prescriptive analytics, an emerging solution category that takes advantage of machine learning to go beyond the insight provided by typical analytics (studying past events) or predictive analytics (forecasting future events). Prescriptive solutions take the next step to suggest what course of action will optimise a process or an outcome.

Some vendors describe prescriptive analytics as offering the "best" course of action for a given set of conditions. That is indeed their goal, but the real power of these solutions isn't that they achieve the "best" result in one go; it’s that they are iterative: They enable positive feedback loops of learning, improvement, more learning, and more improvement over time so their recommendations steadily improve.

Separately, but in parallel, process automation solutions can enable positive feedback loops -- that is, they can become intelligent in their own way -- if they possess certain key capabilities. Perhaps the most important is that they must enable business users to easily automate their own processes by means of self-service, drag-and-drop interfaces that don’t require them to write any code or to gain permission or support from IT.

When given this “self-service” capability, people tend to target the processes that take most of their time, are least efficient or most annoying, and so on. In other words, when they can quickly and easily automate their own processes, they can focus on the ones that will make the biggest difference in their workday.

They also have a strong incentive to pay attention to the results and analyse whether automation has delivered the benefits they were seeking. Often the answer is “partly”: Some benefit has been achieved, but more is possible. In that case, the same capabilities that make these solutions easy to use also enable rapid iteration. They allow people to quickly modify their automated processes to work even better.

Potential for synergy

While machine learning and intelligent process automation are each extremely valuable in their own right, there is tremendous potential for synergy in linking them together. We can do this by instrumenting automated processes to capture data about themselves – such as where, how often, how quickly they execute; the conditions under which they bog down; and how they make decisions -- and then applying machine learning to steadily and automatically improve them.

In this way, we can create systems that work the way people do -- not by following long lists of deterministic rules, but by acting on the equivalent of well-informed “hunches” that become more and more accurate as they are tested in the real world.

There are many promising areas for this type of synergistic solution. One that is useful for illustration is customer interaction and support. The recent explosion of online business has created a need for better tools to manage the customer experience, but maintaining large staffs to provide in-person service is prohibitively expensive. Accordingly, many banks, credit card companies, and wireless carriers (to name just a few of the many relevant industries) have invested in chat services or chatbots that attempt to understand what customers are asking and generate helpful responses.

A growing share of these systems are incorporating machine learning to ingest massive amounts of input and become steadily more accurate in interpreting what customers are saying and asking. Still, many are limited to answering questions such as “what is my balance” or “when is my next payment due.” Often, the next steps exceed what a chatbot can handle and require human interaction, at which point the customer waits in a queue for the “next available representative.” This is both annoying for the customer and expensive for the provider.

A better solution would be to employ sophisticated process automation to deliver satisfying outcomes more quickly and at less expense. For a wireless carrier, this might start with automated analysis of the customer’s history in terms of longevity, timeliness in paying bills, minutes and data used, and the like.

Next steps could include offering rewards for loyalty and/or incentives to renew a contract or purchase additional minutes or data. If a customer is close to being eligible for a new phone, an automated system might recognise that and offer the new phone ahead of schedule as a way to proactively establish greater loyalty. The system could automatically ship the new phone, update the customer record, and manage inventory and replenishment systems.

In most cases today, such processes are handled manually – that is, by hourly workers in call centres. This is not only expensive (labour is often one of the largest, if not the largest, line item in the company’s budget); it is also error prone, especially given the intricacy of most carriers’ service plans. Automated systems can significantly reduce both costs and errors, improving both the bottom line and customer satisfaction almost immediately and increasing customers’ lifetime value to the carrier.

This example is just one of many that illustrate the massively powerful virtuous circle we can create by harnessing machine learning and intelligent process automation. The first step is to instrument the customer’s processes in order to create and capture detailed data about how they execute. Then we apply machine learning to that data in order to identify trends and exceptions and formulate “educated guesses” or hunches about how the underlying processes might be improved. The more data these systems ingest and the more experience they gain in acting on the data, the better job they do at iterating and improving the underlying processes.

All the components of this virtuous circle are evolving quickly, so I expect to see an acceleration next year in the number of enterprises that recognise and take advantage of its power to optimise their businesses.

Alain Gentilhomme, Chief Technology Officer, Nintex
Image Credit: Sergey Nivens / Shutterstock

Alain Gentilhomme is the Chief Technology Officer at Nintex, the recognised global leader in Workflow and Content Automation (WCA), where he brings more than 25 years of experience with product development, product management and technical leadership experience.