The Four-letter Word That’s Key To Making AI Work For Your Business

With SAP’s BTP Innovation Days currently under way, ANZSAP deep-dives into key themes emerging from these sessions to unpack issues that present opportunities or threats to businesses looking to roll out or upgrade scalable business operating systems such as SAP BTP.

Every time AI was mentioned during the Melbourne edition of the BTP Innovation Day, there was a palpable buzz in the room.

The key question on the lips of every attendee, and every business – how to embrace effectively and safely what is considered THE greatest leap forward in business IT since the internet itself?

One common solution quickly emerges in the form of a seemingly simple four-letter word – data.

Susan Tasic-Clark, the Senior Manager of Solutions Consulting for OpenText, told attendees the success of the AI journey of any business could be dictated by the state of its stored data.

With businesses determined to mine every customer or user interaction for valuable insights, and also analysing their own results and performance, they’re aggregating data at a greater rate than ever before.

Yet as Susan warned, as much as 80% of stored data may be unstructured, and up to 62% could be ‘dark’. “That means you don't know the value of it. It also means you don't know the risk,” she said.

“For example, you have people that have worked for your organisation for a long time. Maybe they’ve lodged health data with you, or had a workers compensation situation. All those medical reports or their payment information, the PII (personally identifiable information) in that, you need to know where it is and protect it. That’s a massive consideration.

“You can't do that with dark data. If you don't know what you've got, you can't protect it.”

Established businesses in particular may hold onto significant archives of unstructured data for longer than necessary. “If you've been running your organisation for a long time, think back 10 or 20 years. Things were named terribly, there was no metadata captured. You’ve got no idea what's inside that content,” Susan said.

“The first step is to understand what you've got, and what's the risk of that information? How does it intersect? Where does the information flow from, and flow to, and how can we get better insights in that data? Once you understand what you've got, you're then able to apply the appropriate protection to that content.

“If the data is risky, do you need to keep it? That's another consideration, so many businesses and government agencies we work with, hold on to stuff forever, but do you actually need to keep that? If it’s dark, you can't see into it. You don't know what you've got. You don't know if you can get rid of it. You don't know if it's a duplicate.

“So understanding your overall holdings as you get ready for AI, and looking at a process of what do we need to keep and manage, versus what can we get rid of, becomes really valuable for you.”

Susan said a data management strategy should be considered foundational before applying AI to any use cases within the business.

“By working through some of the data management and foundation, you can really start to solve and move away from the manual effort,” she said. “Then you can look at using your smart people to help them to make better decisions and use some of this AI tooling that's coming up.”