The Xceptor Blog

Making AI-enabled energy operations a reality

06 December 2018

4 minutes read time

Five years ago, a McKinsey report described energy operations as a ‘zoo of different systems.’ So how do you take control of a zoo that has long been running rampant?

It’s a big challenge and one that needs to be tackled in bite-sized pieces. At its most basic level, energy trading is a risk management exercise and, yet, operational risk creeps in by stealth. How so? With a complex myriad of ETRM systems that barely talk to each other, a raft of spreadsheet workarounds and staff running around making it work the best they can with manual fixes, creating their own shortcuts and methods that few people know about. The result is an expensive workshop that is struggling to adapt to the needs of today’s business.

What is needed is an agile trading environment that can cope with the increase in asset classes, the distributed delivery model, the mounds of unstructured data and so on. It also needs to satisfy the Board’s deafening cry to cut costs and deliver ROI.

To do this, a lot of energy firms have taken the first AI-step and turned to robotic process automation (RPA). This is the low hanging fruit of digitisation and is a good place to start if the right expectations are set. Costs can indeed be lowered, and efficiencies captured but, in most cases, it is likely to be 5% of one person plus 15% of another, 14% of another and so on, rather than a straight replace of FTEs.

What RPAs do best is the well-defined, lower value tasks that typically involve structured data rather than enterprise or divisional-wide, more complex processes. This limits the ROI and applications but there are lots of tasks that are suitable i.e. those that are labour-intensive and error prone. Given the scope of these across firms, the robots can pick up their share of the (carefully selected) load.

Look beyond the RPA hype though too. To truly deliver change in your operations and bigger ROI, a fresh approach is also needed. Processes need to be challenged, not just mimicked. Data needs to become the oil of the operation, not the by-product. And we need to harness the power of human intelligence through the deployment of AI.

AI-enabled solutions have the power to automate business processes and operations in ways not previously possible. Complex processes can be optimised, and all types of data, including semi- and unstructured, can be harnessed and enriched. These more intelligent systems can self-learn and provide much needed insights to deliver those nuggets that customers truly value.

As already mentioned, given the myriad of systems, the answer cannot, nor should be, to rip and replace. Rather it is better to deploy solutions that can help extend the life of existing systems, where needed, while travelling along the digitisation journey. By choosing solutions that can deliver both new AI-enabled technologies while bringing together disparate systems, the business can ensure data flows are more connected.

With the business more connected, those more complex, higher value processes can be overhauled and optimised. However, it also seems most AI solutions require data scientists or engineers, and these tend to be scarce and expensive. The key is finding solutions that empower the business users to be able to train the solution. This once again enables the business to drive forward automation and leave the specialist data scientists for their true calling.

It’s time to take control of the zoo, reimagine it. Not just ape it.