More companies are implementing artificial intelligence in their finance operations. Akash Palkhiwala, CFO and COO at software chip maker Qualcomm, talked to me recently about the latest advances the company is making with the technology in that part of the business. Edited excerpts follow.
WSJ: What’s an example of an AI use case you’re still experimenting with in your finance function, but is close to official application?
Palkhiwala: When we update a financial forecast for a period, one of the first things we end up doing once we've locked the numbers in is review changes that have happened. Revenue forecasts year-over-year or quarter-over-quarter. Previous forecast versus new forecast. You want to be able to review what are the things that drove the changes that happened. If the revenue forecast went up, where did the changes happen? What customer, what product, what revenue segment? In the past, it would be research that is done by an analyst to get to a view like that. What we're now seeing is deploying an LLM to be able to look through that same research that an analyst would look at and then publish on that metric.
Rather than a bunch of analysts working on it for a week to do that, we hit a button and we have the first draft done across the board and then analysts are reviewing it to make changes to it and finalize it, versus creating it in the first place. Very, very easy, simple use case that will get deployed at scale in my mind.
WSJ: What’s the potential cost savings of that?
Palkhiwala: There's a time-saving metric that goes with it. We haven't necessarily quantified it directly. Rather than think of it as a cost savings, I'll think of it as productivity improvements, and that allows us to really deploy those resources to a lot of the diversification strategies that we're deploying and allow us to scale the business without additional investment.
WSJ: Is there an AI use case in finance that excites you?
Palkhiwala: Generally, when you think about accounts payables, accounts receivables and manual journal entries, those are all areas that are ripe for disruption with AI. It's something that we are at the front end of investigating, but eventually I think you'll see everyone deploy some form of AI for each of those use cases.
|