Where have your best AI opportunities come from? Have they been driven by business requests, IT initiatives, competitive pressures, or other sources? How has this evolved over time?

I’d echo the same thing. We really focused on experimentation to start, then personal productivity. Now, the use cases are coming out of very specific business processes. Now that you kind of understand the types of things AI is good at, you can start to think about the cases that are best for us to apply it to.

This is self-inflicted, I want to say, on my side. Greg and I have a like-mindedness when it comes to the capability we’d like to take advantage of on behalf of the business as we maintain our position as an IT enabler. Out the gate, no one’s coming to me directly to say, “We would love to do this or that.” I haven’t asked the question yet, but I intend to. Before they ask too many questions, we want to have some preliminary proof already established in a proof of concept. We’re just starting out, and the impetus for this is really my own and Greg’s vision around where we think the opportunities are—for efficiency plays, automation, and just knowing more about large language models and machine learning in general so we can take advantage of things related to IT support, documentation, and all manner of things like that.
Very much at its inception, but driven through a vision I have and the need for injecting this slowly but pretty quickly.

It’s a combination. Our business is pretty savvy, and now that we’re under private equity, our CEO and others sit in forums and councils with broader exposure to what others are doing in the industry with AI. Some good ideas have come from the business asking, “Hey, I heard another company is doing this—can you see if this is something we can do here at US Silica?” There have been a couple of ideas from that, but it’s been mostly an IT push. We’ve taken a conscious approach and tried to put a strategy together so that we don’t fall behind. We’ve done Gartner research, peer connections, and industry polling to narrow down the use cases—identifying the top ten that people have successfully implemented and seen ROI from, and then determining which are applicable to US Silica in the manufacturing or mining industry.
As things have evolved over the last 12 to 24 months, we’re starting to see more ideas come from the business. Now, they’re saying, “I need to be able to do this from a productivity standpoint—can you help me? Can AI help with that?” It definitely started out with IT leading and the business trailing, but now I feel like both are at the table simultaneously, coming up with ideas and picking what we want to spend time prototyping or diving into.