Two of Oracle’s leading executives put forward GenAI ubiquity as a guiding theme of the supplier’s business applications offer in recent customer conference visits to London.
Most recently, Evan Goldberg, executive vice-president of Oracle NetSuite, said that we are still at the dawn of the AI age, “nowhere near noon”.
In an interview with me at SuiteConnect London, he said that while NetSuite “customers have expressed some concerns about GenAI, overall, there is just excitement to use the tools, try them out and experiment with them”.
Goldberg also shared that the supplier’s support staff are reporting time-saving and productivity benefits from using GenAI. “They are getting more done, they can spend time to learn new capabilities in NetSuite. Sometimes they have to edit it, but that’s still better than having to write the whole thing”.
And the bigger picture is, he said, “there is no dumb version of NetSuite”. GenAI everywhere is the supplier’s refrain. “Plus, it’ll get faster and cheaper. We are nowhere near maxing out our datacentre capacity with people using NetSuite”.
He confesses himself surprised by how good GenAI is proving to be, as he found out himself when he had to write a script to support his wife’s company’s use of NetSuite. First he did it manually, then used a GenAI tool – “and it nailed it”.
But he did draw attention to what is a little understood current limitation of generative artificial intelligence – the inability to understand business data as such: numbers.
Searching for a Large Number Model
Golberg again, and it is worth quoting him verbatim here:
“I think the thing that we haven’t really seen yet that is going to be a revolutionary change is analysing business data, and utilizing business data, because that hasn’t been as much of a focus.
“And we’re trying to make it more of a focus [at NetSuite]. We are scouring the landscape for research and small companies that are doing interesting things with business data for forecasting and automation, and the things that we’re concerned about. And we are working with companies like Cohere to think about what is the best place to apply their technology in business, where it is all about numbers. Large Language Models are about language, by definition.
“I’d say we need a Large Number Model. We need models and AI technologies that can deal with sophisticated business data. Harmonising business data is something that we’ve always wanted to be able to look at for our customers in aggregate, and have them put their data into an anonymized pool and make conclusions about it. But every business’s data is slightly different. It seems like AI has a lot of opportunities to harmonise that, and pull out the intelligence from aggregated data.”
Infusing GenAI into Fusion
Steve Miranda, executive vice-president of Oracle Applications product development, shared his thoughts with me on where enterprise IT stands with respect to GenAI, on the eve of Oracle’s CloudWorld London event earlier in 2024.
First he stated that there is a host of AI capability in the Oracle Fusion applications suite already “from supply chain planning to product optimizations. But the focus is on the 50 or so GenAI features we’ve added [in late 2023/early 2024]. These are extremely popular features amongst our customers”. He adduced customer service enhancements, better FAQs, and also using a vector search to compare similarities of service requests and improve those.
The full list of GenAI capabilities that Oracle announced in mid-March 2024 fills out the detail of what Miranda was speaking about in that interview.
In conversation I put to him the fairly common business IT industry thesis that while 2023 was the break out year for Generative AI, 2024 will be the year when it proves its business value. He said: “That seems right. My only hesitation is I fully expect that this year we’ll iterate both on the GenAI engines and the use cases as our customers get their hands on it and get some ideas. And then we adopt, adjust and go forward.
“But I would say, on job posts alone, Gen AI does as good as, probably better job than humans could do at a fraction of the time. So there’s definitely a productivity gain there. I think we’ll get more and more into the value throughout this year and the future.
“It’s all use case driven. There are use cases where Gen AI is very good. There are use cases where other machine learning algorithms are very good. So, one part is finding the right use case. And then it’s [a matter of] the maturity of the models. And again, the accuracy of the models, and how much can you rely on them. And the more accurate it is, and the better it is, you know, the more automation that you get. That’s why we chose the examples that we did, helping with recruiting, helping with answering service questions, doing narrative reporting. GenAI is very good, as the name would imply, at generating language. That is where the big acceleration has been in the last year.”
GenAI in BI
Recent research from TechTarget’s Enterprise Strategy Group, “The State of Analytics and Business Intelligence Platforms” seems to bear out this story of Generative AI failing to have lost its lustre, and making steady progress at companies in North America – though we can always expect fast following in the UK and continental Europe.
In particular, the research, from my colleague Mike Leone, Principal Analyst,
AI Software and Services, Enterprise Strategy Group found that 39% of the organisations surveyed are leveraging Generative AI for analytics and BI, and 41% are using augmented analytics – augmented by machine learning, broadly. It finds usage to be skyrocketing and use cases for GenAI in data management and analytics broadening, from data visualisation (in use now, 40%) through code generation (38%) to summarization of results (35%), and other use cases.
Viewpoint
What is being done with the large language models is astonishing, but Golberg makes a telling point about business data requiring a different, not language-generative, approach. Perhaps it is the scepticism bred of my formation as a journalist, but I am still wary of GenAI boosterism, at least in relation to enterprise IT. It could yet be that GenAI will sediment down to being merely assistive technology, like word processing or spreadsheets, or auto-correction of text messages: barely remarkable. Or maybe we are, as Evan Goldberg suggests, barely at the dawn of a brave new GenAI age?