There is a fundamental road-trip rule: The driver controls the music.

The driver, you see, controls the car. The audio system is part of the car. Ergo, the driver controls the audio system. And she who controls the audio system controls the music. Can a passenger make requests? Yes. Can the passenger occupying that passenger seat be asked to manipulate settings or get us to the right playlist? Sure. But, in the end, the driver controls the music.

Today’s cars have everything from endless variety of curated playlists to digital whoopie cushions that can deploy a range of gas-based melodies, turning the once simple choice of suitable driving music into a potentially overwhelming extravaganza. I rely on my co-pilot’s capacity to listen, synthesize requests around the entire vehicle, make suggestions, but in the end to execute, not distract me from the job at hand.

As a marketer, a co-pilot for this wild ride known as customer engagement is exactly what I and so many of my CMO colleagues have expressed needing for their teams and for themselves. I’m not looking for someone to control the radio…just someone to help recommend some new playlists and time-saving options. What I absolutely don’t need is more complexity…more buttons to push, more apps to sift through and more Teams messages distracting me from the road ahead. I don’t need the promise of perfection, but I would welcome a shortcut or two.

This is where too many early applications of AI (and especially generative AI) have failed the average Marketer. In a time when we didn’t think being a marketer armed with data and technology could be any more complex, we added AI that can spit out commands about what time emails are opened and a dozen new subject lines we should be testing. Far too many AI tools sound like a mystical “easy button” that enter this world fully formed and matured, but in reality they are barely trained to understand language, let alone understanding business or marketing.

When the headlines started screaming how AI could take a marketer’s job, some marketers just laughed. Can any of you remember the last time we asked for something—an email delivery report, a segment analysis, an image for the home page—and we got the exact right thing we wanted the first time we asked for it? While marketers have been curious about the world of possible, many are also weary of all the promises unfulfilled and the implications of applications gone wrong, especially as early examples of AI malcontent have begun to emerge.

This is why I was a bit skeptical when taking a spin through the ongoing advancements Microsoft has been making with its Dynamics 365 Co-Pilots, now available for Sales, Customer Service and Marketing with a preview recently announced for Field Service. Did we really need another tool? Was AI going to make a marketer’s drive that much more enjoyable? Or could this new AI implementation take our eyes off of growth?

Almost immediately I came to an understanding that the CoPilots infused into the Microsoft universe, from Dynamics to Office, were truly there to sit in the passenger seat and work with the driver…not to distract. First things first…CoPilot is not something that a Microsoft Dynamics user HAS to use. Instead, CoPilot is something that can be toggled on. CoPilot also isn’t isolated to a single functional tool which is exceedingly important to Chief Marketing Officers in this age of driving enterprise-wide growth and engagement strategies. CoPilot thrives when cross functional data is unleashed and true insights about the customer, about opportunities and about the work at hand can be accessed.

Let’s take the act of identifying opportunities for growth, a strategy that typically starts with identifying a segment or cohort ripe for profitable engagement. While this sounds simple and like what systems have been able to do before generativeAI, the reality is that the questions marketers learned to ask where limited by their knowledge of the data structures underlying their analytics or engagement systems. Questions were structured around what was known and were actually rather difficult to craft. If we pull back the curtain on the realities of posing these types of queries before large language models and generative AI innovations, marketers would either need to rely on data and analytics teams or craft basic queries from pre-seeded templates. Questions that felt easy to answer demanded prompt engineering expertise and the ability to craft a query in SQL.

Segment building with CoPilot allows a marketer to just ask a question, in their natural pattern of speech, to identify a new segment. But CoPilot can actually take an additional step by including deeper insights to the response, delivering a look-alike segment that also includes details on customer lifetime value or known engagement or contact preferences. It takes "can I see a segment of prospects in the Los Angeles area" to "can I see a segment of prospects within the retail industry with headquarters in the Los Angeles area who have attended webinars and also have an average deal size of over 1 million year to date."

While CoPilot has easily stepped into the spotlight in recent Microsoft Dynamics centric announcements, what caught my eye even more was Microsoft’s move to center both the data AND the engagement layers marketing relies on by bringing their Dynamics 365 Marketing and Dynamics 365 Customer Insights into one solution, now dubbed Dynamics 365 Customer Insights. This shift brings all engagement and experience tools that had existed across the Marketing offerings into a natively integrated, robust customer data platform (CDP) that sits at the ready to ingest, harmonize and normalize customer data and the accelerate to shift to execution and engagement. It intentionally and profoundly connects the stores of data that represent knowledge of the customer with the systems of engagement that power experience delivery.

Yes…in the analogy of who controls the music, what Microsoft envisions here is better integration of car and fuel so that the AI-powered CoPilot has an even greater opportunity to connect to, synthesize and deliver recommendations and actionable intelligence that make the entire driving experience that much better. While the driver is still in total control, there is more readily available information, recommendations, and ambient experiences to make driving easier and perhaps more rewarding.

But for any driver…I mean Marketer…looking out at all these AI innovations, including Microsoft’s CoPilot features in Dynamics Customer Insight, now is the time to raise the question of strategy and deployment. AI, especially generative AI, still requires guiderails to ensure commercial viability and enterprise readiness. It demands a massive corpus of data to train and refine direction and outcomes. It will still require the brilliance, creativity and judgement of the marketer to turn questions and ideas into profitable strategies and executions. The goal of AI for modern marketing is to have a new breed of amazing interns/assistance/copilots, capable of a decision, content and insight velocity that can move us well beyond the limitations of human scale.

My recommendation to Marketers is to start driving with AI in the same way that you started driving a car…you learn, you practice. You don’t necessarily jump into a Ferrari and speed onto the highway on your first ever drive. Instead, you drove mom’s old car in circles in a massive parking lot until you built confidence, knowledge and skill. Start small, but keep scaling and know that not every suggestion will be a win…and not every AI-derived recommendation needs to be followed. These are the early days…the days when the training wheels most definitely stay on.

But these are also the days when you just roll down the window, ask your CoPilot to find something new that you can sing along to, stick your hand out into the wind and just enjoy the ride.