Braze’s AI Marketing Tools Put Campaign Decisions and QA in Beta

Braze’s new AI marketing and customer engagement tools aim to automate work at several points in a campaign, from choosing what to send to checking it for mistakes and handling customer replies. The company announced four products at its Forge conference in Las Vegas, alongside a collaboration with AWS; all four tools were in beta at publication (announcement and product details).

From campaign choices to quality checks

Decisioning Studio Go is the most direct attempt to automate campaign experimentation. Braze says its agents can choose a message variant, timing, day and frequency for each customer, testing more than one billion combinations simultaneously. Marketers set the guardrails and can deploy an agent without relying on technical teams or professional services, according to the announcement (Decisioning Studio Go details).

That scale of testing is a company claim, not a reported independent benchmark. The practical distinction is that marketers define the boundaries while the agent works through many possible combinations. At publication, the product was in beta, with general availability expected on 14 October; the report did not give a year for that date (beta status and expected availability).

Agentic Standards targets a different source of campaign risk: setup and review. It checks campaigns and Canvases for issues with copy, links, personalization, compliance and brand guidelines, then reports pass, warning or fail. Braze says it can work with BrazeAI Operator to implement fixes (Agentic Standards workflow).

Sportsbet’s CRM AI Enablement manager gave a concrete example, saying the system caught missing unsubscribe links and eligibility segments. That illustrates the kind of gap the tool is meant to flag, though one customer’s account does not establish a general error-reduction rate (Sportsbet’s example).

Assistants and conversations beyond the core workflow

Operator Connect links Braze with Claude, ChatGPT, Snowflake Cortex Code and Microsoft Copilot. Teams can use those assistants to analyze performance, develop campaigns, generate content and take actions in Braze; actions remain tied to each user’s Braze permissions (supported assistants and permission model). Stitch’s head of solution delivery said the integration increased speed-to-market by up to 90 percent for its customers. That is a customer-reported maximum, not a general benchmark (Stitch’s reported result).

Conversational Agents extend the workflow to customer-facing conversations across WhatsApp, SMS, RCS and web. Braze says they can answer questions, interpret intent and recommend next steps using brand guidelines, knowledge sources and live customer context (channels and stated capabilities).

The AWS collaboration adds Amazon Bedrock to the BrazeAI Agent Console and expands Braze’s availability through AWS infrastructure and Marketplace, according to Braze (AWS collaboration). The announcement describes a route to use AWS AI capabilities within customer engagement, but does not detail architecture or commercial terms. With every new AI capability still in beta, the rollout status is as important as the breadth of the pitch: the announced workflows are not yet evidence of mature, generally available performance.