Customer engagement platforms crossed a critical threshold on September 30, 2025, when Braze unveiled its BrazeAI product suite at the Forge 2025 conference in Las Vegas. The announcement introduced composable AI agents that make individualized decisions about channel, message, creative, offer, and frequency, representing a fundamental departure from rule-based marketing automation.
The launch signals that AI decisioning has moved from experimental capability to production infrastructure in customer engagement. Brands can now deploy reinforcement learning agents that continuously optimize customer interactions without manual intervention, while maintaining the control and transparency that compliance teams require.
The BrazeAI Product Architecture
Braze released five distinct AI products at Forge, each addressing a different layer of the marketing decisioning stack.
Advertisement
300 × 250
BrazeAI Decisioning Studio reached general availability as the centerpiece of the launch. The system deploys configurable AI agents that use reinforcement learning to personalize countless aspects of each communication. These agents leverage available data, including custom and industry-specific datasets, to identify patterns most likely to maximize rewards and minimize penalties defined by the marketer.
The product represents the integration of OfferFit’s AI Decisioning Engine, which Braze acquired earlier in 2025. Rather than offering a bolted-on feature, Braze embedded the technology as a native platform capability.
BrazeAI Agent Console entered private beta, allowing brands to create custom agents for content generation, data enrichment, and intelligent orchestration. The console supports bring-your-own-LLM configurations from Anthropic, Amazon Bedrock, OpenAI, or Google Gemini, acknowledging that enterprises increasingly demand model flexibility rather than vendor lock-in.
BrazeAI Operator provides unified AI access for campaign building, workflow automation, and real-time insights. Entering limited beta in October 2025 with full beta planned for January 2026, the product integrates with Snowflake Cortex AI for natural-language data queries about campaign performance.
The Model Context Protocol Opens Data Access
Among the more technically significant announcements, Braze introduced the BrazeAI Model Context Protocol (MCP) Server. This enables marketers to connect Braze data directly to AI clients such as Claude Desktop or Cursor, providing conversational access to insights about campaigns, canvases, segments, and custom attributes.
The MCP approach represents a shift in how marketing teams interact with their data. Rather than navigating dashboards or requesting reports, teams can query their marketing data through natural conversation with their preferred AI tools.
Zero-Copy Architecture Removes Data Friction
Beyond AI capabilities, Braze unveiled Zero-copy Canvas Triggers, allowing marketers to activate proprietary data for segmentation and personalization directly from their cloud data warehouse. This eliminates the traditional requirement to replicate data into the engagement platform before using it.
The zero-copy approach addresses one of the persistent friction points in enterprise marketing technology: the delay between data availability and activation. With direct warehouse connections, marketers can act on the freshest possible data without waiting for sync cycles or managing duplicate datasets.
Executive Vision: Composable Intelligence
Bill Magnuson, Braze CEO, framed the launch around the concept of composable intelligence: “This wave of composable intelligence doesn’t just automate tasks; it empowers marketers to orchestrate with greater creativity and precision.”
The “composable” framing is deliberate. Rather than offering a monolithic AI system, Braze structured its AI capabilities as modular components that teams can configure independently. A brand might deploy Decisioning Studio for offer optimization while using the Agent Console for content generation, with each operating according to different parameters and guardrails.
Partnership Ecosystem Validates Enterprise Readiness
The Forge announcements included integration partnerships with Snowflake for data queries and Google Cloud Marketplace for enterprise procurement. Dennis Buchheim from Snowflake noted that the integration “gives marketers the solutions they need to transform raw data into action, accelerating insights.”
Jim Fairweather from Google emphasized the transformative potential: “AI agents play a vital and transformative role in driving business innovation and revolutionize workflows.”
These partnerships signal that BrazeAI is positioned for enterprise-scale deployment rather than mid-market experimentation.
What This Means for the Customer Engagement Category
Braze’s Forge 2025 announcements establish a new baseline for what customer engagement platforms must deliver. The combination of reinforcement learning decisioning, composable agent architecture, model flexibility, and zero-copy data access creates a capability gap that competitors will need to address.
For marketing teams evaluating their engagement technology, the key question is no longer whether their platform supports AI but whether it supports autonomous AI decisioning that improves without manual tuning. The Braze approach, where agents learn from outcomes and adapt their strategies continuously, represents the direction the entire category is moving.
The December 2025 beta of Snowflake Cortex AI integration will further expand these capabilities, enabling natural-language queries about campaign performance and industry benchmarks directly within the Braze environment.
Related: how Databricks and Braze are rewriting customer data infrastructure | campaign creation becoming a prompt
