Process intelligence: Ensuring GenAI speaks the language of your business
The IBM report data and panelist use cases illustrate that GenAI isn’t merely a futuristic concept, but a present-day reality with tangible benefits. However, as Bloehdorn pointed out, the IBM survey also showed that 30% of non-CEO senior executives say their organizations are not ready to adopt GenAI responsibly. A potential hurdle to GenAI, and more specifically LLM, adoption is that the technology doesn’t operate in a vacuum. It needs to speak the language of business to reach its full potential. This is why process intelligence, according to the panel, is critical.
Echoing the words of Filippo Catalano, chief information and digitization officer and Reckitt, who spoke during the Celosphere 2023 keynote, Dudeja said process intelligence offers tremendous opportunity. On the back end, he believes process intelligence will enable GenAI with better data. On the front end, he said it can enable the users to take more data-driven actions.
Dudeja outlined one use case for process-intelligent GenAI that Reckitt is working on–better identification of price mismatches in their procure-to-pay process.
Typically Reckitt does a three-way match, between what’s in their master data, the purchase order (PO) and their invoices. With process intelligence and GenAI, they can expand the process to a four-way match, pulling unstructured data from thousands of legal contracts that come in all different types.
“It's not humanly possible to really extract that information and validate [it] at the transaction level,” Dudeja said. “So when you're looking at a scale, you need that kind of capability from generative AI.”
On the front end, GenAI helps employees better interact with the now four-dimensional data and identify invoices with out-of-sync payment terms or pricing. The technology can then help them take appropriate corrective action and drive working capital improvements.
“If you combine those two together, it's tremendous,” he said. “We have proven the technology, [and] now that it works at both the ends, now we have to see whether it can work at scale.”