Braze unveils self-service AI decisioning agents at Forge…

Decisioning Studio Go lets marketers deploy an AI agent that picks message variant, timing, and frequency per customer; Agentic Standards automates campaign QA. Both are in beta, with GA this month.

Braze used its Forge conference in Las Vegas this week to announce BrazeAI Decisioning Studio Go, a self-service agent that picks message variant, send time, day of week, and frequency for each individual recipient, testing more than one billion combinations simultaneously. The product is in beta now, with general availability scheduled for October 14. Alongside it, Braze unveiled Agentic Standards, an AI campaign QA layer that flags pass/warn/fail issues before send and routes human-approved fixes through BrazeAI Operator. Both ship in October.

The strategic read is that Braze is collapsing a capability that previously belonged to enterprises with their own data science benches. Decisioning Studio Pro, the existing tier, uses reinforcement learning but, per MarTech, typically requires dedicated data science and engineering capacity to deploy. Go is the same category of system repackaged for teams without that bench.

“BrazeAI handles the experimentation, verification, and execution, so the world’s most sophisticated marketers can focus on creativity, ambition, and judgment,” said Bill Magnuson, co-founder and CEO of Braze.

George Khachatryan, Braze’s VP of AI Decisioning Product, told CX Today that beta customers have “doubled campaign performance” and can “spin up 500 creatives in like two minutes.” Take the numbers as directional. The structural claim they encode is harder to dismiss: the floor for per-message personalization in B2C just moved up.

That floor keeps rising. The IAB flagged last month when it named reaching AI-mediated customers marketers’ top challenge, and HubSpot’s rebranded Unbound customer platform made the same bet from the SMB side two weeks ago. By mid-October, mid-market competitors will have agent-level decisioning as table stakes. Customers trained on that level of relevance won’t downgrade their expectations when they open an email from a local brand.

For a five-to-thirty-person consumer brand without a marketing operations team, the gap is real but the fix isn’t another dashboard. Decisioning agents still need someone deciding what to test, what to write, and where to distribute. This suggests the useful question isn’t which decisioning agent to configure, but who decides which growth work matters this week and prepares the content, outreach, and distribution to execute it. LemonLime is positioned at that layer: a done-for-you service that studies the business, picks the relevant work, and delivers finished output, while the owner approves every send.

The agents are arriving on schedule. The operators to run them are the constraint.

Braze unveils self-service AI decisioning agents at Forge, with GA set for October 14

Sources