AI has stopped being something marketers try and became something they run on. That was the clearest signal from Forge, Braze’s annual conference in Las Vegas.
The ideas on stage weren’t radically new. The difference is that they work now, well enough to get all the way from strategy through execution. That opens up a lot. It also makes it easier than ever to be exactly like everyone else.
Most of what Braze announced is live today. Decisioning Studio Go, Conversational Agents and Operator Connect are in beta, and Agentic Standards lands this month.
AI moved from the sidebar to the workflow
For two years, AI in marketing platforms mostly meant helpers. A subject line suggestion here, a job aid there. Useful, optional, easy to ignore.
This year, nearly everyone at Forge described the same shift. AI is becoming the main way teams do the work, end to end. A marketer can explore a strategy, draft the journey, generate and test the content, and launch it, with AI doing most of the assembly at each step.
The releases that make it concrete:
For two years, AI in marketing platforms mostly meant helpers. A subject line suggestion here, a job aid there. Useful, optional, easy to ignore.
This year, nearly everyone at Forge described the same shift. AI is becoming the main way teams do the work, end to end. A marketer can explore a strategy, draft the journey, generate and test the content, and launch it, with AI doing most of the assembly at each step.
The releases that make it concrete:
- BrazeAI Operator. Builds a full Canvas journey from a plain-language description. A first draft lands in minutes, so teams spend their time on strategy instead of assembling steps.
- Content Optimizer. Tests hundreds of subject line, copy and CTA combinations and shifts traffic to the winners. Motorway saw 114% more clicks and 37% more valuations.
- Conversational Agents (beta). Two-way dialogue on WhatsApp, SMS, RCS and web, drawing on brand guidelines and live customer profiles.
- Operator Connect (beta) and the Braze MCP Server. Operator now works inside Claude, ChatGPT and Cursor, and other AI tools connect to Braze through one URL. Teams can build and manage Canvases from the AI workspace they already use.
The payoff is real. Small teams can act big. Big teams can move fast. Fewer handoffs to IT, analytics or an agency just to get something live.Braze also made a clear strategic choice. While many vendors keep adding clouds to cover every corner of marketing,
Braze is going deeper on the channels and core product its customers already run on. For brands, that means a platform that keeps getting better at the job they bought it for.
The race to the middle
One of the key risks presented by all this advancement is something that feels like a “great race to the middle.” When every brand uses the same tools to optimize for the same metrics, everyone lands in the same place. Or, as Braze CEO Bill Magnuson put it, “everything is f*cking boring.”
We’ve seen this movie. Streaming gave us more TV than anyone could watch, tuned to every taste. The prestige era gave way to an endless scroll of shows built to be good enough. Optimizing for engagement produced a lot of content but very little anyone remembers.
Marketing is next in line. Efficiency is about to be table stakes and the brands that win will use the time AI gives back to make work people notice.
While the platform can’t deliver highly creative breakthrough strategies on its own (yet?), Braze did release a set of guardrails alongside the agents, and they matter for keeping automated work accurate and on-brand:
- Agentic Standards. Checks plain-language QA rules, like brand voice, disclaimers and tagging, before launch. AI-built campaigns get the same review as any other.
- Message Prioritization. Ranks message categories once and reschedules lower-priority sends when they collide. Customers stop getting three messages at once, and the most important one goes first.
- Canvas Threshold Alerts. Flags journeys whose send volume drops below normal, so you catch a broken journey before customers or leadership do.
- Messaging Observability Dashboard. Shows why a message was sent, held or dropped.
One-to-one personalization is finally within reach
Marketers, agencies and platforms have chased one-to-one personalization for more than 20 years. Mostly it has lived in pitch decks. Recently, and in particular at Forge, it felt achievable.
It still takes work: the right data foundation, the right tools, the right strategy and a program to run it. But brands are doing it now. For example, our client Intuit TurboTax is getting very good at putting the right offer and message in front of the right customer, on the right channel, at the right moment.
Several Forge releases push in this direction:
- BrazeAI Decisioning Studio Go (beta). Self-serve AI agents pick send time, creative and messaging for each customer, without a data science team.
- Accounts. Groups related users under a shared account, so billing and product-update messages reach the right people. Built for B2B and multi-user brands.
- Surveys. Drag-and-drop in-app and landing-page surveys. Detractors route to a support Canvas and promoters to a referral Canvas, so feedback triggers action instead of sitting in a report.
- Shopify Markets, Segment Sync and Banners. Localized language, pricing and consent sync, market audiences, and no-code on-site banners. Regional commerce programs run from one place.
Expect the gap to widen fast between brands that can do this and brands that can’t.
None of it works without the data foundation
Braze spent a lot of stage time on its partnerships with Snowflake, Databricks and AWS. Part of that story is marketplace access, which makes Braze easier to buy. Nice, but it’s the smaller story.
The bigger one: cloud data platforms are where brands keep the volume of customer data that one-to-one personalization runs on. Getting that data into Braze, and getting engagement data back out, is what makes AI decisioning work. An agent choosing the next best message is only as good as what it knows about the customer. Braze’s data releases are built for that loop:
- Databricks OpenSharing. Two-way sharing of Braze engagement data with the Lakehouse. Engagement data feeds your models, and model output flows back into messaging.
- User Profile Streaming. Keeps the warehouse current with Braze attributes like loyalty tier or churn risk, so analytics and Braze work from the same profile.
- Cloud Data Ingestion. Adds Google Cloud Storage and Azure Blob as sources, with a no-code attribute mapper. More places to pull data from, with less engineering work.
This is where Bold Orange lives. We build data foundations in Snowflake, from data engineering to analytics, so the data is clean and usable. And we run Braze and platforms like it, so that data turns into messages customers act on. Finally, we bring the marketing know-how to decide what’s worth saying in the first place.
Here’s What We Recommend Brands Do Next
Check your data foundation.
Can your warehouse feed Braze the attributes your personalization depends on? Does engagement data flow back? If the answer is “sort of,” start there. Everything else depends on it.
Start experimenting.
The tools are live today. Pick one build feature, like Operator or Content Optimizer, and pair it with a guardrail like Message Prioritization or Agentic Standards. Run a real pilot this quarter, and test the beta tools before rolling them out broadly.
Protect the unexpected.
Let AI handle the assembly. Keep people on the ideas. When everyone can optimize, a message that surprises is worth more than it was a year ago.
It’s an exciting moment for our clients, and for us. If your Braze setup is underused, or you’re not sure which of these releases fits your program, let’s talk.
