Kevin, I'd say this. We did with our data science group, we did a deep dive, looking back for the last 3 years. We analyzed the results from our roles that are vulnerable based on World Economic Forum, which are the customer service, the coders, the lower level operational level positions, and we looked at that in very granular detail and found that it performed no differently than the rest. Further, NFIB did their own study. And 98% of their constituents said, AI had no impact on their number of employees. And now you look at the Stanford, Harvard, Yale and basically say to the extent there's been an impact, it's on early career, entry-level people. Well, guess what? That's not our business. Our clients won't pay us to get for them early career, entry-level people because they can do that themselves. They don't need us for that. And so to the extent that's where there's an impact has been. It's easier to understand or it certainly confirms our own internal studies that there's no impact from that. And so you then say, look, okay, well, then why has the industry been down for 3 years? And that's where I come back to, let's talk about churn. Well, let's look at JOLTS. In October of '22, there were 6 million people hired, and there were 4 million people that quit. You roll that forward to August of '25 there were 5 million hires in the United States, and they were 3 million quits. And that's a huge difference. There's a lot less churn. That churn plus the job growth that has taken place in the United States have been concentrated in government, clinical health care and leisure and hospitality. And those are not big consumers of contract temporary help for the industry and particularly for Robert Half. And so I feel about as confident as I feel with anything that AI has not contributed to what has happened so far, so far, either for the industry or for Robert Half. And instead, it's about clients as they focused on their cumulative inflation issue as they worry about all of these forecasts of recession, they become more cautious and juxtaposed against -- they want to keep their full-time staff, what's their first lever to control their cost? Fewer contractors. So the industry now is in year 3 of companies trying to keep, retain their full-time staff, control their cost and it's been primarily at the expense of their contractor usage. I think all those dots connect, but you don't need AI as an impact to connect them. It's less churn, it's narrow growth, not that applicable to the industry and/or Robert Half.