What happens to brands when personal AI agents seek the best deals on behalf of their consumer owners?

Into a cocktail shaker, mix today's focus on privacy, advances in machine learning and voice response systems, and consumers' overwhelming desire for convenience. Shake vigorously. The result is an ecosystem of personally owned and controlled AI agents that predict needs, shop for solutions, negotiate terms, and place orders.

To the public, this is a refreshing beverage that eases their troubles and is the digital assistant they've been promised since Star Trek in the 1960's. To marketers, this is a bitter concoction that instantly generates an excruciating headache.

What's a brand manager to do?

1. Become and Remain Top of Mind

While mindshare has been the remit of every brand manager from the start, tomorrow things go beyond complex and sophisticated, and go straight to convoluted.

Today, if I ask Alexa for a package of AAA batteries, Amazon branded batteries appear at my door. If you have done your job well, I will ask for Duracell batteries by name.

Tomorrow, my AI agent will predict my smoke detectors will need fresh batteries and it go shopping on my behalf. It will find the best products according to published product testing, consumer research, and public opinion. It will query online stores that have a reliable, negotiate the best price, delivery, and return policies, and place the order.

If you have done your job well, my agent will place a higher score on your batteries. You can offer Amazon a higher margin in return for market share. You can create a supply chain agent to work with Amazon's API. You can offer kick-backs to executives scraping together enough for their next Tesla. But you will not beat all competitors until you have so firmly placed "Duracell" in my lexicon that my agent knows I will reject any other brand.

You will have to do what you do now, only more so.

2. Accept the Inevitable

This does sound like so much science fiction, but then so did the idea of an index-card sized object in your pocket that can navigate strange city streets, make restaurant reservations, predict the weather, and make telephone calls. Just as an Internet enabled iPad may have sounded far-fetched in 1990, personal agents are going to happen. Get used to the idea now.

3. Learn About the Tech

Don Draper would have understood psychology and market research. Twenty years on, he would have needed database and statistics smarts. Today, he'd be studying machine learning. You do not need to be a Data Scientist to use machine learning, just as you don't need to be a programmer to advertise on Facebook. But you must understand the concepts and limitations to keep up with the times.

4. Buy, Don't Build

You cannot afford or even find a Data Scientist and a Data Scientist doesn't know your business. You know your customers, marketplace, and competition. You can use AI features big companies are embedding in their marketing platforms as well as the new tools start-ups are creating daily.

5. Rest Assured, You Have A Job

Brand managers will always be needed for three reasons. Your first job is to determine what problem to solve or question to answer. The machine will simply sit there, waiting for you it give it a task. Second, you must discern which data sets the machine should consider when coming up with answers. It only considers what you feed it. Third, you will gauge whether the output is reasonable. The machine does statistics; it has no common sense, reason, or imagination.

6. Start Now

Try it. Test it. Play with it. Start a pilot program to see what works and what does not. Do not wait, thinking you can be a fast follower. The more you try, the more you learn and that is exactly how machine learning works.

The robots are coming. They are coming to help you do your job and to make your job much, much harder. We live in interesting times.

Jim Sterne produces the Marketing Analytics Summit and is Board Chair Emeritus of the Digital Analytics Association. His twelfth book is Artificial Intelligence for Marketing.

Originally posted on Branders Magazine