The word “influencer” has gained a new definition over the last few years as Instagram, Twitter, and Youtube have given rise to a new generation of charismatic talking heads who can sway the opinions millions of subscribers. In simple terms, if people see an influencer they like do something, those people are likely to do that too.

Some influenced behaviors are easy and cheap; things like running every day, or eating certain foods. But making a profit has become a common practice for many influencers. If they show off a new pair of running shoes, suddenly their followers now have reason to want running shoes from the same brand. It only makes sense that influencers would seek money for this practice, especially when brands and advertisers are eager to have influencers showcase their products. A win-win.

Enter Influential, a Los Angeles-based company that began using IBM’s Watson Ecosystem, an artificial intelligence program that scans natural language used on social media, to pair businesses with influencers in 2016, says Influential CEO Ryan Detert.

Influential uses the Watson A.I. to interpret language used online companies, influencers, and followers of both. By analyzing text from platforms like Facebook, Instagram, and Twitter, Influential is able to make optimized matches between companies and influencers: finding parties that use similar language, and whose followers use similar language, to decide which companies and which influencers should team up for advertising.

Influential CTO Piotr Tomasik says that while the A.I. is an immensely helpful tool, they are far from a completely automated company. “We’re only touching the surface on what we can do,” he says. “Job-replacement fear? I think we have a pretty long way to go to be, like, exact and perfect.”