There's a bit of a sotto voce debate going on among agencies right now about the best organizational model for exploiting the potential of artificial intelligence for their clients. Some say AI and machine learning are so new and impactful, you've got to focus a team and fence it off in a separate division or subsidiary so that it can explore their potential free of the politics and mundane, everyday influences of current mainstream work. Others say AI and ML need to be integrated into everything you do for each client.

I first met Amy Ingram through an email she sent to arrange a meeting with the CEO of a New York AI company. My exchanges with her were peppered with phrases like "I'm sorry," "Have a nice weekend," "Thank you." I had no idea Amy Ingram was a bot.

What if you knew the concerns that were uppermost in your customer's mind, right now? Or, which of the folks in your Eloqua database are most intensively researching the topic of the white paper you're about to blast out - this week? Or, best of all: Which Fortune 1000 companies are intent about buying something you have on offer - ASAP? I recently learned these things actually are knowable, today, given just a few machine-learning algorithms and a database the size of a small outer planet.

If there are roughly 400,000 journalists, bloggers, and other potential influencers publishing in the English-speaking world, wouldn't it be great if you could send that press release about your lovingly brewed craft beer only to those who actually cared? Last week, I learned of a U.K. start-up using AI machine learning technology to solve this problem. Trimaldi Ltd., just shy of nine months old, recently announced an AI-backed press release distribution service that uses machine learning algorithms to send a marketer's release only to relevant influencers.