Published on September 29, 2025/Last edited on September 29, 2025/7 min read
AI decisioning is revolutionizing the way marketers create 1:1 personalization across the customer lifecycle. Built on reinforcement learning, AI decisioning agents act as the brain that sits atop your marketing tech stack, learning how each unique customer interacts with your brand. Marketers choose the business metric they want to maximize, and AI agents personalize every aspect of a campaign based on unique customer profiles.
The goal is to realize the dream of true 1:1 personalization, building more lasting relationships with customers while impacting the bottom line. But, without the right resources in place to set you up for success, this type of marketing AI can also be challenging, complex, and labor intensive to implement and use effectively.
Some AI decisioning technologies advertise themselves as out-of-the-box, self-serve tools that are designed to run autonomously. That all sounds great, but for the foreseeable future, some level of human oversight is going to be necessary.
If expert services are not available via the technology provider themselves, the burden could be on your company to make sure the AI acts as intended and delivers on the goals you’ve set. That’s a tall order, and something most brands don’t have the resources to take on by themselves.
At Braze, we work hand in hand with our customers to build custom AI decisioning systems, offering human-in-the-loop solutions that accelerate and augment the value of AI decisioning via expert services teams with years of experience. We provide our customers with much more than decisioning agents, we provide true partnership and expertise—forward-deployed engineers and AI experts who will help you make the most of our technology and bring your vision to life.
As the Head of AI Decisioning Deployment at Braze I’ve seen firsthand what can happen when AI decisioning agents are deployed without the proper guardrails and expertise. I often describe AI as one of the smartest people who’s ever lived, but someone so smart they may lack common sense. If left unchecked, the technology can single-mindedly attempt to help brands achieve a stated goal, without taking other business realities into consideration.
For instance, if your company is offering discounts and your campaign is set to maximize clicks or conversions, the AI could potentially do whatever it takes to achieve that goal at the cost of giving away too much value.
This happened when one of our customers developed an in-house model to calculate customer lifetime value (CLTV) and wanted to use AI decisioning to optimize the CLTV for each customer. During testing, our team discovered the AI decisioning agent decided to meet this goal by offering one month of free services to 80% of customers, rather than applying the special offer more selectively.
It turned out that the AI decisioning agent was behaving as designed, but there was a flaw in the customer’s CLTV model that led to it looking like the one month of free services offered was disproportionately valuable. It took our experts analyzing the in-house model the customer had developed to identify a problem that might otherwise have gone undetected if the company had gone live with this pilot with a self-service technology that didn’t offer any technical support or oversight.
Decisioning platforms backed by forward-deployed engineers and AI experts recognize that the metrics that matter the most and the kinds of data that drive the most meaningful actions vary greatly across industries. In contrast, self-serve platforms often miss the mark.
For example, a quick-service restaurant may obsess over profit margin and turn to insights like customers’ past purchases, topping choices, and dietary preferences. In contrast, a financial services brand might be more interested in cultivating high-value referrals and personalizing offers based on customers’ spend data, repayment activity, and credit scores. The objectives, data sets, and regulatory restrictions will be very different across these 2 industries and an AI decisioning system that can’t be heavily customized will fail to capture value. However, customers typically don’t have the training or resources to do this sort of deep customization, which is why it’s necessary to pair flexible AI decisioning with expert services to fully capture value.
AI decisioning that’s designed with humans in the loop will help integrate the right data and act on it in ways that create value for different business types, recognizing that offering anything more than even a few dollars in promotional discounts could cut into QSR brands’ profit margins. In contrast, for a financial services brand,offering hundreds of dollars to banking customers for referring new customers who have the potential to bring in thousands of dollars in business is likely to be a very sensible strategy.
Using AI decisioning without technical support can feel like getting the keys to a self-driving car from a dealer who offers no more guidance than wishing you luck and sending you on your way. With Braze, we not only give you the keys to the car, we put you and the car through driver’s ed, make sure you both pass, and we’re on standby to offer roadside assistance every step of the way.
While other providers in AI decisioning tend to take a one-size-fits-all approach when it comes to implementation, we partner with you in thinking through how AI can best solve your challenges. Our white glove services team will become deeply grounded in your business needs and strategic priorities to help you leverage AI decisioning to meet your brand’s unique use cases.
Our teams can bring general guidance—such as the insight that it’s more effective to send emails and SMS at off-hour times, such as 12:51 p.m. instead of 12:00 p.m.—as well as industry-specific knowledge, such as understanding how demand patterns change due to seasonality, which is something that can be tricky to train agents to understand, but that we’ve done plenty of times.
Many organizations may not have the internal data engineering resources necessary to get AI decisioning programs off the ground—our team can fill in the gaps. With Braze you get education, enablement, and support at a fraction of the cost it would take to build and maintain an in-house team of data scientists, while also getting a diversity of thought and experience across industries that an in-house team would lack.
Various team members across your organization are likely to care about different metrics. Our team can help you navigate these internal politics and relationships to ensure you’re able to communicate the ROI our system is delivering and how it aligns to these different goals.
Setting AI decisioning systems up improperly from the start could increase costs in the long run, especially if the AI decisioning system operates in ways that cuts into profit margins—you’re leaving money on the table. Conversely, building the expertise in house could cost time and money, resources many companies don’t have.
While off-the-shelf, plug-and-play providers will require teams to stick to pre-configured integrations, Braze can work with you to bring the data sources you need to guide decision making.
Sign up for regular updates from Braze.