Key points
- Suntory Oceania is investing in Databricks and Copilot Pro while exploring native agentic options for its SAP core, digital and IT director Matt Dixon said at the SAP NOW AI Tour in Sydney.
- The company has moved about 90 percent of its datasets, roughly 40 in total, onto Databricks as a centralised data foundation over the past two years.
- Dixon envisions using SAP to automate demand sensing, with an agent eventually raising stock transfer orders that a person would only need to review and approve.
(L-R) Paul Dearlove (DXC) and Matt Dixon (Suntory Oceania).
Suntory Oceania is bolstering its AI capabilities with investments in Databricks and Copilot Pro, while at the same time exploring native agentic options for its SAP core.
Digital and IT director Matt Dixon told the SAP NOW AI Tour in Sydney that the beverage maker had started a concerted effort to prepare for AI about two years ago.
At the time, Suntory Oceania was working through a major transformation connected to a business decision to insource alcohol manufacturing, and to bring its alcohol and non-alcohol beverage businesses together.
A three-year transformation ensued to stand up the required plant, corporate facilities, and related infrastructure, including a suite of “new digital tools”, which it had assistance from DXC in standing up.
“We had a go-live date of July 1 2025 that was non-negotiable. We couldn’t miss that date – and we hit it,” Dixon said.
While the organisation was focused on the main transformation, AI started to make waves in the enterprise space, to the point that it became a “parallel stream” of work to the main transformation.
“Partway through the [transformation] program, we were actually looking at how we can make sure that we’re AI ready in the future?” he said.
“Whilst we couldn’t change the scope of what we were working on, there were things that we could do to start getting AI ready.”
AI adoption has so far manifested in three different ways at Suntory Oceania.
First, the company has adopted Databricks as a centralised data foundation, from which it can build out AI use cases.
“Two years ago, we [started] moving off our legacy data platforms and up into Databricks,” Dixon said.
“We’ve spent the last two years doing that. About 90 percent of our datasets are now up in Databricks – something like 40 datasets are up there now.
“We’re 1792616105 able to start leveraging that environment [and] we’re able to start building tools on top of that.
“The next step for us is to see how we can actually start injecting that into some of our [customer experience] tools.”
AI for personal productivity is another use case for the technology.
“We’ve been seeding and scaling Copilot Pro, making sure that’s getting into the hands of the people who are going to actually make the most use of it,” Dixon said.
In terms of license numbers, Dixon said the company is buying “a couple hundred licenses this year, [and] a few more hundred next year.”
“We’ll keep on scaling that up whilst also making sure the whole business has got access to the standard version of Copilot,” he added.
Suntory Oceania also has interest in agentic AI, but while Dixon said he had received a number of vendor approaches with varying options, he wanted to first explore options that are native to the company’s core SAP environment.
“Really we need to make sure we’ve looked at what is available in our core first,” he said.
“That’s where we are right now, doing that evaluation.”
Demand sensing
Also within its core SAP environment, Dixon envisioned using the software to make aspects of its operations more autonomous and data-driven.
An example he raised was around demand sensing: essentially using data signals in near real-time to drive supply chain adjustments.
“What can I see in the future? Something like we see an uptick [in demand], some sort of activity that we hadn’t anticipated, like a heat wave [driving] an uptick in consumption of our non-alc and alc beverages,” he said.
“That sort of demand sensing at the moment is all manual. People have got to recut the plan, do inventory optimisation manually, and then they’d have to raise a stock transfer order to rebalance to meet the demand.
“Now, not if [but] when we have that demand signal optimised, and we’re seeing that sellthrough from our customers that that can be updating our demand planning algorithms, I can see an agent in the future that actually has that demand sensing, looks at the inventory optimisation that needs to happen… in the [SAP] platform [and] raises a stock transfer order to get the product where it needs to be to meet the demand.
“That sort of demand sensing is never going to be real-time, but it’s going be a lot more immediate than it currently is.”
He added that the only human interaction in the process might be having a person review the logic the agent has used to raise the stock transfer order, and OKing it.
“They’re the kinds of things that I see in the future for us.”
