Customer Engagement platform provider Braze has announced several new additions to its AI portfolio, continuing its effort to bring one-to-one decisioning to more customer journeys. These new additions focus on giving Marketers more control over AI-driven decisioning and agentic automation, while also ensuring clear governance over what agents do.

Last year, Braze announced BrazeAI Decisioning Studio Pro, an agentic decisioning/personalization layer that uses reinforcement learning to determine the right channel, message, offer, timing, and frequency for individual customers. Reinforcement learning is a form of machine learning that continually learns what works best for the customer and adapts the experience accordingly. Agents in the Pro version of the Studio are highly customizable and configured by forward-deployed data scientists.

Braze has now announced an entry tier for Decisioning Studio called Decisioning Studio Go. It’s for marketing teams that want to replace static rules and manual AB testing. Marketers use Go to set up and configure AI decisioning themselves. Right now, Go makes one-to-one decisions for content, timing, and frequency for email journeys.

Give Decisioning Studio Go an audience segment and a set of variants to work with, including subject lines, calls to action (CTAs), images, send days and times, and the decisioning agent will pick the best ones for each individual in the segment. The agent can test over a billion combinations simultaneously to find the best combination.

Decisioning Studio Pro is for large companies with complex personalization requirements, but for companies that are looking to scale more basic personalized email journeys that don’t require data scientists or professional services, Go is the best place to start.

Marketers want to ensure that everything their agentic agents do meets the team's standards. But putting a human in the loop to double or triple-check everything won’t scale. The question is, how do they give AI more autonomy without losing control? For Braze customers, it’s through new Agentic Standards.

Agentic Standards are a new approach to quality assurance. It is an AI-powered review process that checks campaign setup, copy, links, personalization logic, and compliance requirements against brand requirements the marketing team defines in the Braze platform. Every check is either a pass, warning, or failure.

These standards run at build time when a marketer is doing QA before launching a campaign or Canvas (Canvas is Braze’s visual journey builder). For example, the standards could verify that a personalization tag is correctly formatted and that a backup value is provided if the data used to populate the tag is missing. Every check and fix is logged, ensuring a full audit is available.

With an audit complete, the next step is to fix any issues that surfaced, and that’s where BrazeAI Operator comes in. Operator is like an AI assistant that marketers use to get help with activities such as setting up campaigns, fixing QA issues, providing data analysis and reporting, and automating repetitive tasks (you could compare Operator to HubSpot’s Breeze Assistant).

According to Braze, Operator can be set to auto-approve or have every choice approved manually. Most marketing teams will fall somewhere between automating simple tasks and carefully reviewing more important activities.

For marketers who prefer to work in AI tools like ChatGPT, Claude, and Microsoft Copilot, Braze has extended Operator’s functionality through Operator Connect. This is a Model Context Protocol (MCP) Server that also enables Braze partners to build Operator-powered capabilities into their own products.

Use Operator Connect in ChatGPT, for example, to ask Braze questions and get answers, like the results of a recent campaign. Or ask it to set up a new email campaign with specific rules. Partners can connect their own agents and platforms to work with Braze capabilities.

The point is, Marketers don’t need to switch between tools. Before, they had to research and ideate in an AI tool, then go to the Braze platform to implement a campaign or run a report. Now, they can do it all within the AI tool. As more and more marketers turn to AI tools as their go-to workspace, platforms like Braze need to ensure they can connect with those tools. The MCP Server is a must-have for every software provider right now.

The other big product announcement from Braze is the introduction of Conversational Agents. These agents, which work for WhatsApp, short message service (SMS), rich communication services (RCS), and web, are context-aware, meaning they retain the context of the past 48 hours of messages on that channel.

Currently, the agents are channel-specific, so they don’t carry context across channels (something Twilio does), but it’s likely on the roadmap; it’s a key requirement for delivering better customer experiences across channels.

Conversational agents work with brand-specific guidelines and guardrails, along with deep knowledge bases, to ensure they provide the best experience for customers.

In the past year or so, Braze has been on a run developing new agentic capabilities for its engagement platform. It’s doing what every Martech technology must do - become AI-native.

But as it introduces new agentic capabilities, are marketing teams evolving the way they work to truly take advantage of them, or are they still trying to fit traditional Marketing processes into tools that can do so much more?

Sometimes, I think the bigger question is: how are Marketing teams evolving their strategy and processes?