Very interesting. I didn't know that. Do you think that because they rely on their own data centers, it might put them at a disadvantage as AI becomes more integrated into these products? It seems like if you're making a big bet on AI and embedding it in your products, there's conventional wisdom suggesting you need to adopt hyperscalers and run things on their platform. It's hard to manage that in-house. What do you think?

I think it poses a real threat. Paycom has brought in AI personnel to develop internal large language models. They're trying to keep up because they're not positioned to move as fast as others in the space. They're not connected to the larger cloud-based infrastructure where AI is being pushed. If someone wants to do that, they have to build it themselves. This is something they need to figure out, especially if AI functionality becomes more prevalent in the front end of these platforms. HCM and Payroll have been slow adopters of AI due to the sensitive nature of the data, like pay, direct deposit, or Social Security. The tolerance for inaccuracy is very low. However, I think it's only a matter of time before full-blown automation through AI is seen in many core HCM vendors.