The technology sector has always been prone to hype and promises of silver bullets. Marketers get ahead of engineers and evangelize over the promise of the Next Big Thing. Conferences dedicated to the new hot topic mushroom everywhere. Journalists write features about it while analysts plan white papers. Venture capital firms scour the planet for startups specializing in it. Firms tell you that they’re all about the Next Big Thing.
Play it again, Sam? Well, you sure have heard it all before. It’s happened with RFID, with service-oriented architecture, cloud, blockchain, 5G and dozens of other hyped technologies, and today it’s happening with artificial intelligence, or AI.
AI is great, and it’s certainly one of the most promising areas for changing the future of computing and business automation – even if we sometimes need to remind ourselves that it has been around as a concept for about half a century. The temptation when there is so much buzz around a topic is to pretend that it is some sort of universal panacea with broad applicability. But, with apologies for being a party-pooper, let’s be straight here: it’s not needed for 80 percent of what an ERP does.
Explore related questions
“To a man with a hammer, everything looks like a nail,” goes the old phrase and it is certainly apposite here. And it’s important to say why that is, because too often we still look at IT challenges from the context of examining what new tools can give us rather than a much more important, high-level concern that all CIOs should keep as their first question: what are today’s business challenges and opportunities that we need to address?
It would be very easy to succumb to the AI hype around automation in enterprise environments and join in as companies seeking to raise funds or raise profiles tend to do. Smart CIOs should certainly be scanning AI developments for a range of reasons and use cases but, in ERP at least, they would be better advised to spend more time pondering what elements are involved in a business process and joining the dots accordingly. So, I want to propose a simple formula here: in ERP, Integration > AI.
Only connect
Integration may not score you tens of millions of venture capital, win you a book deal, nor make you the subject of media headlines too often. The fact remains though that it’s a crucially important (and lamentably neglected) aspect of ERP success.
Perhaps understandably, people rush to shiny new things and consider how they can be applied to their world. AI (and its subset machine learning) have been fawningly profiled as the next great disruptors for business technology, but in today’s enterprise applications their relevance is an outlier. That’s not to say we don’t see value and promise in areas such as project resourcing or workforce planning where an ML algorithm could, say, identify repeated patterns and potentially make accurate predictions as to the success or otherwise of various approaches. But the dirty secret of AI is that we can do a lot of this stuff anyway today, as using tried and tested algorithms and development tools that are already here.
Success in ERP is about integrating relevant data from across the software stack to enable rapid processing pipelines and prevent silos from emerging. You wouldn’t have much of a chance of success in building a new football stadium if you included material costs and availability and conducted research into location and land, but had omitted to factor in costs of people with the right skills and when they could do the work.
Good ERP systems effectively force users to consider practicalities and they help to surface anomalies. And if the integration of various component elements is strong, then you have excellent prospects for business infrastructure success. But if core elements of criteria such as time and billing are not well integrated, then you have a very big problem indeed.
Think first, act later
The goal therefore should be to look across the organization to understand all the data elements necessary for a given business process and connect them seamlessly. And if you can automate so that, for example, the ERP system knows a person is traveling in a certain country and can therefore more easily understand and process local-currency expense claims, then you start to give time back to users and to finance departments, and both sets of constituents should acknowledge the smoothness and convenience of the user experience.
All too often, however, gaps emerge because processes are created without the input of all relevant parties. This leads to systems that are unworkable, suboptimal, require retooling and frustrate users who are unable to perform key tasks. Of course, organizations don’t stand still, so even a system that was perfect for its moment in time will need to be re-examined later. As such we need to do a great deal of work upfront, but also we need to use microservices so that when needs change over time, we can adapt systems.
All organizations today need to reduce operational costs and weed out manual processes, especially given an economic outlook that points us directly to creating efficiencies wherever possible. Automation should be front and center all the way through a process. In HR, for example, that process would extend from recruitment and onboarding, to skills identification and scope for career development, all the way through to end-of-tenure agreements and the creation of alumni networks. But AI is not a synonym for automation, and most of these scenarios are very doable without AI.
AI is great, but…
AI is not pixie dust and even where it is highly useful, remember that AI will still require the precepts of clean, accurate, centralized data. We are very far from The Singularity and AI today won’t necessarily be able to answer highly complex questions with multiple variables because, well, algorithms just don’t work that way. If we haven’t built a platform for scoping and understanding a process, AI will offer us the square root of nothing.
The best CIOs should always maintain a healthy dose of skepticism and AI is no exception to the rule. It’s wonderful that we live in an age of innovation and every technology leader needs to dedicate some time to understanding incoming waves. But they should also always be asking a selfish question that leads to focus; ‘what’s in it for us?’ Technologies are not blanket phenomena: they tend to be good for some things and less good for others. CIOs need to weigh their relevance rather than get caught up in the hype or the attractions of being associated with the latest trend.
So, once again, it’s not my intention to denigrate the very real present and future uses of AI, only to state my belief that recognizing the importance of interconnectivity and integration will take you a very long way without it. Most of us still use only a fraction of the capabilities in our ERP systems so we should be wary of putting the AI cart before the horse. If we examine the promise of AI with this clear headed approach, we stand much more chance of reaping real benefits from our time and our tools.
Claus Jepsen is chief technology officer of Unit4.
