VML MAP

𝗠𝗖𝗣 𝗳𝗼𝗿 𝗠𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴 𝗖𝗹𝗼𝘂𝗱 𝗘𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁: 𝗔𝗰𝗲 𝗼𝗿 𝗗𝗼𝘂𝗯𝗹𝗲 𝗙𝗮𝘂𝗹𝘁? Salesforce just dropped the #MCPServer for SFMC, and I’m split between “this changes everything” and “isn’t this just another API wrapper?” Let me try to call the score fairly: 1️⃣ 𝗙𝗶𝗿𝘀𝘁 𝗦𝗲𝘁: 𝗪𝗵𝗮𝘁 𝘁𝗵𝗲 𝗠𝗖𝗣 𝗳𝗼𝗿 𝗦𝗙𝗠𝗖 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗘𝗻𝗮𝗯𝗹𝗲𝘀. The MCP (𝘔𝘰𝘥𝘦𝘭 𝘊𝘰𝘯𝘵𝘦𝘹𝘵 𝘗𝘳𝘰𝘵𝘰𝘤𝘰𝘭, but you knew that) server exposes an impressive set of tools for a hybrid workforce to interact with journeys, create SQL, manage data extensions, insert contacts and much more (note the “dry run” that allows us to validate destructive actions before execution. Not a bad idea.) Advantage, MCP. 2️⃣ 𝗦𝗲𝗰𝗼𝗻𝗱 𝗦𝗲𝘁: 𝗕𝘂𝘁 𝗪𝗮𝗶𝘁, 𝗛𝗮𝘃𝗲𝗻’𝘁 𝗪𝗲 𝗔𝗹𝗿𝗲𝗮𝗱𝘆 𝗕𝘂𝗶𝗹𝘁 𝗧𝗵𝗶𝘀? Every SFMC agency has developed their own toolkit to operationalise SFMC. 𝘞𝘩𝘺 𝘰𝘯 𝘦𝘢𝘳𝘵𝘩 𝘴𝘩𝘰𝘶𝘭𝘥 𝘸𝘦 𝘣𝘪𝘯 𝘪𝘵 𝘢𝘭𝘭 ���𝘯𝘥 𝘫𝘶𝘮𝘱 𝘰𝘯 𝘵𝘩𝘦 𝘔𝘊𝘗 𝘫𝘶𝘴𝘵 𝘣𝘦𝘤𝘢𝘶𝘴𝘦 𝘪𝘵’𝘴 𝘴𝘩𝘪𝘯𝘺?  And let’s not forget about the legitimate concerns that come with AI agents operating on production instances : destructive tools are only one hallucination away from mass-generating a new genre of (very human) headaches. 30-love. 3️⃣ 𝗧𝗶𝗲-𝗕𝗿𝗲𝗮𝗸: 𝗪𝗵𝗲𝗿𝗲 𝘁𝗵𝗲 𝗥𝗲𝗮𝗹 𝗔𝗱𝗱𝗲𝗱 𝗩𝗮𝗹𝘂𝗲 𝗟𝗶𝘃𝗲𝘀 So why is it genuinely interesting? 𝗕𝗲𝗰𝗮𝘂𝘀𝗲 𝘁𝗵𝗲 𝗟𝗟𝗠-𝗻𝗮𝘁𝗶𝘃𝗲 𝗶𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲 𝗰𝗵𝗮𝗻𝗴𝗲𝘀 𝘄𝗵𝗼 𝗰𝗮𝗻 𝗼𝗽𝗲𝗿𝗮𝘁𝗲 𝘁𝗵𝗲 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺. Today, building and executing against SFMC APIs requires developers. You need to understand authentication flows, REST payloads, journey JSON schemas.  The MCP shifts the interface from drag and drop to (really) declarative: 𝘏𝘦𝘳𝘦 𝘪𝘴 𝘸𝘩𝘢𝘵 𝘐 𝘸𝘢𝘯𝘵, 𝘨𝘰 𝘢𝘯𝘥 𝘥𝘰 𝘪𝘵. It’s not about SFMC’s capabilities, it’s about 𝗵𝘂𝗺𝗮𝗻 𝗲𝗺𝗽𝗼𝘄𝗲𝗿𝗺𝗲𝗻𝘁. 𝗠𝗮𝘁𝗰𝗵 𝗣𝗼𝗶𝗻𝘁: G𝗮𝗺𝗲 𝗰𝗵𝗮𝗻𝗴𝗲𝗿 𝗼𝗿 𝗴𝗶𝗺𝗺𝗶𝗰𝗸? 𝗡𝗲𝗶𝘁𝗵𝗲𝗿! 𝗜𝘁’𝘀 𝗻𝗲𝘄 𝗯𝗮𝗹𝗹𝘀. 🎾🎾 In my view, the MCP for SFMC is less about what it does today and more about the muscle memory it builds for tomorrow’s architecture. For now ? The game doesn’t change. The rules don’t change. The court doesn’t change. But the balls are fresher, livelier, and the serve has a little more bite. So... Tennis, anyone? 𝗪𝗵𝗮𝘁’𝘀 𝘆𝗼𝘂𝗿 𝗿𝗲𝗮𝗱 𝗼𝗻 𝘁𝗵𝗲 𝗠𝗖𝗣 𝗳𝗼𝗿 𝗦𝗙𝗠𝗖? 𝗔𝗿𝗲 𝘆𝗼𝘂 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘁𝗲𝘀𝘁𝗶𝗻𝗴 𝗶𝘁 𝗶𝗻 𝗮 𝘀𝗮𝗻𝗱𝗯𝗼𝘅, 𝗼𝗿 𝗵𝗼𝗹𝗱𝗶𝗻𝗴 𝗼𝗳𝗳 𝘂𝗻𝘁𝗶𝗹 𝘁𝗵𝗲 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 𝘀𝘁𝗼𝗿𝘆 𝗶𝘀 𝗰𝗹𝗲𝗮𝗿𝗲𝗿? 𝗗𝗿𝗼𝗽 𝘆𝗼𝘂𝗿 𝘁𝗵𝗼𝘂𝗴𝗵𝘁𝘀 𝗯𝗲𝗹𝗼𝘄. #SalesforceMarketingCloud #SFMC #MCP #MarketingAutomation #MarTech #AIAgents Salesforce SFMC #DigitalMarketing #MarketingTechnology

For a long time, I measured my Braze expertise by how much I could do inside it. Webhooks to transform data. Custom attribute calculations built in the platform. If Braze could handle it, that's where it got done. More in Braze meant better.... Or at least that's what I thought! I've moved on from that. Not because Braze is limited or it isn't capable - it's one of the most capable platforms I've worked with. But because I misunderstood what made it powerful. And I'm sure many others are making this same mistake! Braze is a composable platform. It's built to connect - to sit at the centre of a stack and orchestrate what happens across it. That's not a gap in the platform. That's the point of it. And once you start treating it that way, the solutions you can build become genuinely different. The webhook example is the one I come back to most. You can chain webhooks to manipulate and transform data inside Braze. Technically, it works. But when you have a dedicated transformation tool upstream that handles that job with proper logging, error handling, and maintainability - and you offload it there - Braze gets to do something better with that data. It gets to act on it. Intelligently. At scale. In real time. That's the real shift in mindset. Not doing less in Braze. Freeing Braze up to do more of what it's actually built for. The same logic runs through every layer of the stack. Clean, well-structured data flowing into Braze from the right places doesn't constrain what you can build — it's what makes the complex stuff possible. The personalisation, the real-time decisioning, the kind of customer experiences that actually move metrics. Being a true Braze expert isn't about knowing everything Braze can do. It's knowing how to architect the correct solution - inside Braze, around it, or with the right tools powering it. #Braze #MarTech #CRMStrategy #ComposableStack #CustomerEngagement

Just two weeks after Adobe Summit, it’s already clear something bigger is happening… and it’s not just “another set of releases.” At VML MAP, we’ve been connecting the dots across everything that was announced — and honestly, the shift is hard to ignore. What stood out the most this year wasn’t a single feature, but a pattern: 👉 Agentic AI everywhere, meeting practitioners where they are 👉 Open MCP-based integrations 👉 Adobe moving from CX Management to defining the CX fabric that AI build on In our recent internal Adobe Guild session "Unpacking the Summit", one thing really clicked for me: we’re no longer just configuring platforms — we’re starting to orchestrate ecosystems. The idea of an “AI coworker” capable of planning, connecting data across tools, and executing actions is not a concept anymore — it’s becoming product reality. And even more interesting: Adobe is opening up. We’re moving from “everything inside the Adobe box” to a world where you can plug Adobe into whatever stack you’re working with — Copilot, ChatGPT, Claude, or beyond. So what does this mean in practice? For us at VML MAP, it means one thing: 👉 we’re actively evolving our practice to become more agentic by design Designing agents + skills, not just workflows & journeys Thinking in terms of orchestration, not just execution Exploring how to combine Adobe + external AI ecosystems Reimagining use cases that used to be “too complex” or “too manual” And maybe the most exciting part? This is one of those rare moments where experience resets a bit — everyone is learning again. The playbook is being rewritten. So yes, two weeks later, we’re still unpacking… But we’re also starting to build 🚀 Curious to hear how others are interpreting the “agentic shift” after Summit.

𝗔𝗴𝗲𝗻𝘁𝗳𝗼𝗿𝗰𝗲 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀 𝟭/𝟱 𝘞𝘩𝘢𝘵 𝘪𝘴 𝘈𝘨𝘦𝘯𝘵𝘧𝘰𝘳𝘤𝘦 𝘖𝘱𝘦𝘳𝘢𝘵𝘪𝘰𝘯𝘴 (𝘢𝘯𝘥 𝘸𝘩𝘢𝘵 𝘪𝘵 𝘪𝘴𝘯'𝘵). On April 29, 2026, 𝗦𝗮𝗹𝗲𝘀𝗳𝗼𝗿𝗰𝗲 𝗹𝗮𝘂𝗻𝗰𝗵𝗲𝗱 𝗔𝗴𝗲𝗻𝘁𝗳𝗼𝗿𝗰𝗲 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀, claiming it cuts back-office cycle times by 50–70%, eliminate 80% of manual data entry, and do it without replacing existing systems. 𝘉𝘰𝘭𝘥. Before we look at Agentforce Operations with our tech lens, let's establish exactly what this product is, and also... 𝗪𝗵𝗮𝘁 𝗶𝘁 𝗶𝘀 𝗻𝗼𝘁: It is not Salesforce Flow. Flow routes work to people and automates decisions within Salesforce. It is not Einstein Copilot, which is a prompt-driven assistant that requires a human to initiate every action. It is not Agent Albert, Agentforce Operations does not learn and replicate what we do. And it is not a rebranded version of the existing Agentforce front-office platform. Phew. 𝗧𝗵𝗲𝗻, 𝘄𝗵𝗮𝘁 𝙞𝙨 𝙞𝙩? Agentforce Operations (built on Regrello, a supply chain automation company Salesforce acquired and fully integrated on October 2025) is an 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝗯𝗮𝗰𝗸-𝗼𝗳𝗳𝗶𝗰𝗲 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗹𝗮𝘆𝗲𝗿. It takes a 𝗽𝗿𝗼𝗰𝗲𝘀𝘀 (procurement, onboarding, invoice auditing, compliance validation) 𝗰𝗼𝗻𝘃𝗲𝗿𝘁𝘀 𝗶𝘁 𝗶𝗻𝘁𝗼 𝗮 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝗯𝗹𝘂𝗲𝗽𝗿𝗶𝗻𝘁, and 𝗱𝗲𝗽𝗹𝗼𝘆𝘀 𝘀𝗽𝗲𝗰𝗶𝗮𝗹𝗶𝘇𝗲𝗱 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 to complete the work end to end, across disconnected systems, 𝘢𝘥𝘥𝘪𝘯𝘨 𝘩𝘶𝘮𝘢𝘯 𝘪𝘯𝘵𝘦𝘳𝘷𝘦𝘯𝘵𝘪𝘰𝘯 𝘰𝘯𝘭𝘺 𝘸𝘩𝘦𝘯 𝘪𝘵'𝘴 𝘯𝘦𝘤𝘦𝘴𝘴𝘢𝘳𝘺. It 𝗰𝗼𝗻𝗻𝗲𝗰𝘁𝘀 𝘁𝗼 𝘁𝗵𝗲 𝗯𝗿𝗼𝗮𝗱𝗲𝗿 𝗔𝗴𝗲𝗻𝘁𝗳𝗼𝗿𝗰𝗲 𝟯𝟲𝟬 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺: Data Cloud for grounding, MuleSoft for system connectivity, Atlas Reasoning Engine for agent decision-making. It is not a standalone product. It is an extension of the agentic enterprise architecture Salesforce has been building since late 2024. 𝗧𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝗶𝘁 𝘀𝗼𝗹𝘃𝗲𝘀 𝗶𝘀 𝘁𝗵𝗲 𝗴𝗮𝗽 𝗯𝗲𝘁𝘄𝗲𝗲𝗻 𝗮 𝗳𝗮𝘀𝘁 𝗳𝗿𝗼𝗻𝘁-𝗼𝗳𝗳𝗶𝗰𝗲 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 𝗮𝗻𝗱 𝘁𝗵𝗲 𝘀𝗹𝗼𝘄, 𝗳𝗿𝗮𝗴𝗺𝗲𝗻𝘁𝗲𝗱, 𝗺𝗮𝗻𝘂𝗮𝗹 𝗯𝗮𝗰𝗸-𝗼𝗳𝗳𝗶𝗰𝗲 𝗽𝗿𝗼𝗰𝗲𝘀𝘀𝗲𝘀 𝘁𝗵𝗮𝘁 𝗶𝗻𝗲𝘃𝗶𝘁𝗮𝗯𝗹𝘆 𝗯𝗿𝗲𝗮𝗸 𝗶𝘁. At least, that's what Marc Benioff claims, and he's probably not wrong. OK. We get the idea.  𝗕𝘂𝘁 𝗶𝗻 𝗿𝗲𝗮𝗹 𝗹𝗶𝗳𝗲, 𝗵𝗼𝘄 𝗱𝗼𝗲𝘀 𝗶𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗯𝗲𝗵𝗮𝘃𝗲?  Well, I'm glad you asked because that's what we will be looking at tomorrow. 𝗙𝗼𝗹𝗹𝗼𝘄 𝗺𝗲 𝗳𝗼𝗿 𝗺𝗼𝗿𝗲 𝗶𝗻𝘀𝗽𝗼 𝗮𝗯𝗼𝘂𝘁 𝗔𝗴𝗲𝗻𝘁𝗳𝗼𝗿𝗰𝗲 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀 #Salesforce #AgentforceOperations #AgenticAI #SalesforceArchitect #DigitalTransformation Salesforce VML MAP

Everyone's asking the wrong question about BrazeAI Agent Console. "Should we be using agents?" is not it. The question is: what do you need agents to do? I've spent time hands-on in the beta. The way I think about it, there are three fundamentally different jobs you can put agents to work on — and picking the wrong one is how you end up with underwhelming results and a very sceptical stakeholder. Job 1: Content generation. The agent reads real-time context such as purchase history, journey stage, catalogue data, and writes personalised copy at send time. Not a template, not static variants. Fresh content for each individual, at scale. Output quality is almost entirely down to your instructions and your data. Messy data model, vague brief = generic output. The agent doesn't fill in what you haven't given it. Worth being clear: agents don't learn from outcomes. No feedback loop, no mechanism to track whether that subject line actually converted. If you need AI that continuously optimises against a KPI, you want Decisioning Studio — a different product, a different engine. Job 2: Decisioning. Instead of sprawling if/else branches in Canvas, you deploy an agent to evaluate a user's full context and route them in real time. Lead scoring, churn risk, sentiment-based branching - without hardcoding a single rule. The logic lives in natural language now. Easier to update, easier to explain to a client, easier to fix when the brief changes at 4pm on a Friday. Same caveat, LLM-based reasoning, not reinforcement learning. Know what you're using. Job 3: Data enrichment. Honestly, this wasn't the one I was most excited about going in. It's become the one I'd tell most teams to start with. Agents parse unstructured input: survey responses, free-text fills, SMS replies and return clean structured outputs. Sentiment scores, variables and catalogue enrichments. Cleaner data means better audiences. Better audiences mean everything downstream works harder. Here's the thing I didn't see coming until I was actually building with it. A single agent is useful. But build an ecosystem, each agent doing one job, feeding context to the next - and the impact gets very real. A single, fit-for-purpose agent replaces canvases that previously needed obscene amounts of Liquid and dozens of steps. Weeks of building become days. One more thing: you're not locked into BrazeAuto. Connect your own provider — OpenAI, Anthropic, Bedrock. Depending on your setup, that could mean pointing an agent at a model already configured for your business, not a generic LLM out of the box. Most people don't realise that's an option. These features are new, but if this is the starting point, what comes next is genuinely exciting. Marketers have always been constrained by what Liquid could express and how long it took to build. Agents start to remove both. That changes things. Which job are you starting with — or are you already composing them? #BrazeAI #CRMStrategy #MarTech #AIAgents

Meet the faces of AI DAY 🚀 On May 6, 2026, CBS AI Academy is hosting AI DAY at Copenhagen Business School, bringing together speakers, founders, jury members, panelists, and moderators for an afternoon and evening focused on applied AI, real business value, responsible implementation, and the next generation of AI builders. 📍 Venue: Ovnhallen, Porcelænshaven, Copenhagen Business School 🗓️ Date: May 6, 2026 🏢 Partner Expo: from 14:00 🎤 Main Event: from 15:30 We are excited to welcome: Speakers 🎙️ Arnold Boon, University Director, Copenhagen Business School Tanneguy Salavert, Founder, Youth Tech Forum Mette Beck-Nielsen, CEO & Managing Director, Digital Dogme Kamille Overgaard Björnsson, Project Lead, Digital Dogme Sannah Brieghel, Founder, Skiftr Thomas Foege, Founder, Skiftr Villads Leth, CEO & Future Unicorn Founder, Wilmo.ai Kasper Juul, CEO & Co-Founder, Kin AI Jury ⚖️ Julie Mørch Nadelmann, Director Enterprise Services Sales, Microsoft Denmark & Iceland Pernille Bernth, AI Lead, Fellowmind Henrik Fabrin, Head of AI, Danish Industry Kim Rants, CEO & Founder, Alice Jerrit Jungmann, VC, People Ventures Jacob Sørensen, Senior Operating Officer & Head of Data Consulting, VML MAP Mikkel Riis, Cloud & AI Executive, VENZO Villads Leth, CEO & Future Unicorn Founder, Wilmo.ai Kasper Juul, CEO & Co-Founder, Kin AI Panel discussion 💬 Christoffer Mc Carthy Mors, Director of Customer Success, Microsoft Katrine Bach, Co-Founder & CEO, Connected Women in AI Suzanne Lauritzen, Founder ExitWomen, Investor & Associate Professor David Pontoppidan, Head of Business AI, Nordics & Baltics, SAP Tina Dyhr Dupont, Director of Customers and Insights, Velliv Moderators ✨ Claus Raasted, Director, The College of Extraordinary Experiences Olivia Linder Tabel, Director of AI at Bech-Bruun, Speaker & Educator, OliviaKnowsAI From keynote talks and panel discussions to live MVP pitches and jury feedback, AI DAY is about connecting ambitious students with the people and organizations shaping the future of AI in business 🤝 Sign up here: https://lnkd.in/eiB8nh4f

Discounts alone won't cut it anymore. 🚨 We interviewed Paul Jempson, Strategy Director at VML MAP, for our Global Customer Loyalty Report 2026. And he had some pretty interesting things to say: Customers want recognition for *everything*: purchases, engagement, data sharing. They also expect brands to anticipate their needs and deliver seamless experiences across every channel. The winners? They’re not just running loyalty programs, they’re building loyalty ecosystems! Paul also called out something big: loyalty is becoming an *income driver*, not just a marketing tool. Powered by AI, predictive personalization, and authentic partnerships. Our report data backs the rising interest in loyalty and AI: • 🤖 51.4% of marketers now use AI in loyalty (up from 37.1% last year) • 💰 51.5% of the marketing budget now goes to CRM & loyalty And you know what? We're here for it! For more expert insights, dive into our Report and watch the full webinar on Antavo’s website. These insights were also used as a source for our Global Customer Loyalty Report 2026. #GlobalCustomerLoyaltyReport2026 #loyaltyprograms #loyaltyinnovation #loyaltytechnology #customerloyalty #loyaltytrend #customerexprience #customerengagement