What we're talking about this month — where humans fit into a world full of agentic workflows, spring cleaning tools for your Braze instance, the best Braze features from recent releases, our top resources, and more! ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏    ͏      Welcome to the April edition of Cutting Edge: the monthly newsletter for marketers who use Braze! Spring has a way of making everything feel possible again. The sun is shining. Plants start growing. And if you've been heads-down all Q1, April is a natural moment to get some fresh air and ask: are we set up for what's coming next? Because a lot is coming. AI isn't a future-state conversation anymore. Braze’s recent AI releases make it available mid-journey, directing decisions inside your Canvases, generating content at send time, and running quietly in the background while your team focuses elsewhere. The question is no longer whether you should engage with it. It's whether your team is positioned to direct it. This month, we're covering what that shift can actually look like in practice — from figuring out where humans best fit into agentic workflows to making the most of Braze's latest releases — plus tips and resources to help you spring forward. What we're talking about this month: 🤖 In the world of AI agents, where do humans fit in? 📌 Braze tips & tricks 🛠️ The most relevant updates from Braze’s recent releases 🪡 Resources you can use to hone your craft STITCH IN TIME What Does "Human-in-the-Loop" Actually Mean Now? This is a question I've been sitting with lately: in a world of AI agents, where does the human fit in? Because the honest answer is: it's changing faster than most marketing teams have noticed. Inside Braze, agents are already operating mid-journey. Agent Console fires LLM intelligence directly within Canvas flows, generating personalised content, enriching catalogs, adapting in real time. Decisioning Studio optimises send-time and channel without a human making that call. At Stitch, we're already building workflows for cross-platform agentic processes. Agents that monitor performance in Braze and surface insights into Slack before anyone asks. Agents that produce HTML against a Figma brief and QA that email content in every conceivable way, keeping project management tools updated, all without a human in the middle. The quality of output is superhuman and the operational layer of marketing runs continuously. So where does the human fit? The traditional marketing role (operator/executor) is shifting to architect. The marketer who will thrive isn't the one doing the most tasks; it's the one who designs the system that does the task, sets the guardrails, and knows when to step in. That's not a diminished role. It's a more leveraged one. But it does require a deliberate answer to the question most teams haven't asked yet: which decisions still need a human, and which ones are we just slowing down by insisting on one? - Dan Elman Senior Solution Architect & Fawlty Towers Fanatic PIN IT Braze tips & tricks 📌 Burning through webhook credits faster than expected? Check your Canvases for webhook steps calling Braze's own /users/track endpoint. That's an easy one to clean up. User Update steps handle the same profile writes — attributes, events, purchases — without touching your quota. And they're more powerful than most teams realize. The JSON editor gives you full Liquid support. Conditional logic, dynamic values, and event properties all work the same way it would in a webhook body. Keep your webhook credits for use cases that actually require them. To audit: pull up your active Canvases and look for webhook steps pointing at your Braze REST endpoint. If the payload is just writing back to user profiles, swap it for a User Update step. - Tatum Lynch Consultant Solution Architect & Mega Metallica Fan PATCHWORK Decoding Braze product updates 🛠️ Canvas Context Variables & Canvas Context Step Context variables are temporary, journey-scoped variables that are created and redefined each time a user enters a Canvas. They persist only within that Canvas entry, meaning multiple active Canvas states can coexist for the same user with independent data. Canvas Context component is a dedicated Canvas step type that lets you create or update context variables for a user mid-journey (complements the Context Variables feature above). Why it matters: Context variables solves one of the more painful limitations in multi-entry Canvas design — the inability to maintain isolated state across concurrent journeys. Previously, you'd have to rely on custom attributes to simulate this, which created messy data hygiene and cross-contamination risks. What it looks like in real life: A travel brand running a post-booking nurture Canvas can use a context variable to store booked_destination and departure_date at entry. If the same user books a second trip and re-enters the Canvas, their second journey carries its own context — so messaging for Trip A and Trip B stays accurate and independent, with no custom attribute juggling required. Braze Agents in Agent Console AI-powered agents you build and configure inside Braze to generate content, make decisions, and enrich data in real time. You define the agent's purpose and guardrails; once live, it can generate personalized copy, make real-time decisions, or update catalog fields. Use a Braze-provided LLM or connect your own AI model providers (such as OpenAI, Anthropic, or Google Gemini). Why it matters: This moves Braze from a messaging execution platform toward an autonomous personalization engine. Agents reduce the manual effort of content production and enable true 1:1 personalization at scale — without requiring an external AI integration. What it looks like in real life: A QSR brand could configure an agent that generates personalized re-engagement offer copy dynamically based on a user's last order, preferred location, and time of day — eliminating the need to manually write dozens of content variants and instead letting the agent produce them at send time within defined guardrails (e.g., always include a CTA, never mention competitor brands). Cloud Data Ingestion (CDI) Sources [Early Access] A redesigned CDI UI that separates sources (the connection/credential config) from syncs (the actual data pipeline jobs). One source can now power many syncs. Why it matters: Previously, every sync required its own full source configuration — duplicative and error-prone, especially for teams managing many tables or schemas. This makes CDI significantly more maintainable at scale. What it looks like in real life: A retailer with a Snowflake warehouse syncing 10 different product catalog tables can now define a single Snowflake source connection and fan out 10 separate syncs from it — meaning credential updates or endpoint changes only need to happen in one place. CSV Pre-Import Validation & Error Reporting Before a CSV user import fully processes, Braze can now scan the file and generate a report distinguishing errors (rows that will fail entirely) from warnings (rows that will partially succeed). Post-import failure reports are also available. Why it matters: Bad CSV imports have historically been silent failures or required manual investigation after the fact. This gives ops and CRM teams confidence before committing a file, and a clear audit trail after. What it looks like in real life: A CRM team preparing a one-time upload of 50,000 suppression records can validate the file first, catch 200 rows with malformed email addresses before import, fix them, and re-upload — rather than discovering the issue post-send via deliverability anomalies. - Gabi Weimerskirch Senior Solution Architect & Piano Teacher STITCH ASSESSMENTS Spring clean your martech stack 🧼 We have two assessments we are offering right now to help you get Braze and Databricks in order and drive more out of your martech stack. Braze Lifecycle Assessment: A full audit of Braze (every Canvas, every campaign) to find opportunities to reduce spend, optimize lifecycle journeys, review technical architecture and data foundations, and team workflows. Marketing Data Assessment: A full audit of your Databricks <> Braze pipeline to identify data you're consuming but not leveraging (which is hurting your ROI), and what data you're not leveraging that you should be. Both investments start at $5k and are part AI-led/part human-led. Both require a half-day workshop with your team. And both assessments generate roadmaps that are yours to keep. Reach out via our website to learn more or reply to this email. Let's talk assessments > HONE YOUR CRAFT Our top resources this month 🪡 🤖 Agent Console, on demand. Our most recent Braze Innovation Series session explored the rapidly evolving landscape of AI tools in Braze, and it’s now available to watch on demand! From a broad overview of Braze’s AI capabilities to a deep dive into the newly released Agent Console, we focused on what’s available today, how it actually works, and where it can meaningfully improve customer experiences. Watch Agent Console webinar > ✍ To CDP, or not CDP? If your team is evaluating a CDP, read this first. We break down why most brands already have the data infrastructure they need, and why the real gap is often activation rather than unification. Spoiler: a data warehouse + Braze connection might be all you need. Read CDP article > 📚 What we're reading/watching Why martech consolidation business cases fall short - Martech.org Measuring AI ROI in Marketing Operations - Knak LOOSE ENDS This month's VIP (very important pet) 🐾 Stitch, 830 Massachusetts Ave Floor 3, Suite 9, Indianapolis, IN 46204, USA Unsubscribe Manage Preferences