This transcript has been lightly edited for clarity.
What differentiates agentic AI, and why is it critical in the race to rewire operations?
Many companies are already adopting gen AI and agentic AI. The underlying technology is all the same. What makes agentic AI especially good for operations, and companies that focus on operations, is that it is all about doing things. What makes agentic AI different is that it will do the thing for you.
If you think about a company's process and operations, a lot of these tasks require judgment, inputs, and then activities happen—you have to make certain determinations. In the age of gen AI, you still require a human to make those decisions or changes in manufacturing, procurement, or product development. But in the world of agentic AI, that part can be automated.
The value at stake for agentic AI is significantly larger. We're seeing manufacturing or product development processes shrink in lead time by 20 or 30 percent. The critical part to deliver impact is where you want to deploy these agents and how they work together to realize the expected impact.
How mature is agentic AI in operations today, and where are the greatest opportunities?
The maturity of agentic AI in operations is, honestly, at a very early stage. I think there are a lot of root causes to that. The number one reason: companies typically see this as a big change to how they do things right now, and business operations are very complicated by nature. However, the potential and the value are absolutely there.
We see a lot of early movers starting to see significant impact, particularly in areas such as procurement, where we already have traditional AI, such as spend cubes or different breakdowns of cost.
If you take that and you apply negotiation strategy, you can build agents that help to understand different negotiation tactics and come up with recommendations to help our supply managers negotiate more effectively. This can bring improvements of 5 to 10 percent in terms of bottom-line procurement costs.
In manufacturing, deciding how to translate initial product design into configuration of a manufacturing line and how to break down the different steps into a manufacturing assembly process can be done in days or weeks with agentic AI.
Where do companies get stuck when scaling agentic AI, and how can they break through?
There's a lot of different research out there, but the consensus is that out of 100 companies that attempt such transformations, 90 percent don't see real financial benefit.
Although the future potential of agentic AI is very promising, most operations companies are still facing a lot of hurdles. Unless you have top-down, leadership buy-in, most companies are in a wait-and-see mode. They want to see what is actually proven, what has been done elsewhere, before they adopt it internally.
That poses two risks: Number one: you are going to be left behind with a higher cost base than your competitors. Number two: this is not a situation where it's going to fix itself. The core gap in most organizations is a talent one. Because if you're not willing to take the steps to understand what can be done with agentic AI, you're not going to build a team behind it that can actually make it happen.
Successful implementation of agentic AI, at the end of the day, still comes down to:
- knowing where the impact is and where you want to embed agents into your processes
- reimagining those processes so that humans and agents collaborate together to maximize the impact
- having a clear execution engine and governance so that all of this can be deployed effectively
