We Were at Agentforce World Tour London 2026: Our Insights

Events
29 Jun 2026
If you were at ExCeL London on June 18th, you already know the energy was something else. Thousands of Salesforce customers, partners, and practitioners crammed into one of London’s biggest event venues for what Salesforce billed as the Agentic Enterprise moment. Walking the floor, sitting through sessions, and talking to people at the booths, it’s a different experience from watching the recap on Salesforce+. So here’s my honest take on the day: what was announced, what it means, and what I think actually matters.
The Big Picture: AI Agents Are Now Expected to Do Real Work
The keynote opened with a clear message: the conversation has moved on from “is AI useful?” to “how do you run your organization with agents doing actual work alongside your people?” That shift in framing felt real to me, not marketing fluff, but a genuine change in the conversations happening at every booth and session I visited.
What I found most useful was a framework shared on the main stage is five things every Agentic Enterprise needs: context, work, agency, engagement, and insight. It’s the clearest articulation I’ve seen of what Salesforce is actually building toward.
What Was Actually Announced
Headless 360 and the Claude Connection: Ask Salesforce Without Leaving the Chat
This was announced a couple of months back, but World Tour London was really the moment it landed for a UK audience, and I think it’s the announcement with the most immediate practical implications for Salesforce users.
Headless 360 is a complete rethink of how the Salesforce platform exposes itself. The whole thing is now available through APIs, Model Context Protocol tools, and CLI commands, so agents, developers, and other systems can work through Salesforce without a traditional Salesforce screen in front of them.
But here’s the thing that got me most excited about it: what this means when you connect Salesforce to Claude via MCP. In plain terms, you can now ask Claude a question and get a real answer pulled directly from your Salesforce org. No logging in. No running reports. No waiting for someone on the ops team to pull the data. A sales rep can ask “what’s the status of our top 10 open opportunities this quarter?” and get an actual answer, right here, in the chat.
And it goes further than read-only queries. With the right MCP setup, sales teams can register new opportunities, update deal stages, log activities, and close deals: all through a conversation with Claude, without touching the Salesforce UI. Think about what that means day to day:
● Account managers can ask “which of my accounts haven’t had any activity in 60 days?” and get a list they can act on immediately.
● Marketing can pull campaign response data and segment lists on the fly, without building a report.
● Customer success teams can check renewal dates, flag at-risk accounts, and trigger follow-up tasks. Just by asking.
● Event teams can query registrations, check capacity, and update attendee records in a conversation.
The new experience layer that comes with Headless 360 also extends native Salesforce interactions to Slack, voice, and WhatsApp, so it’s not just Claude. Any surface, any channel, same Salesforce data underneath.
There is one catch, and this is important: all of this is only as good as the data sitting in your Salesforce org. If your opportunity stages are inconsistent, your contact records are outdated, or your account data hasn’t been maintained, Claude will surface exactly that mess, just faster. Headless 360 doesn’t fix data quality. It amplifies whatever’s already there. So before you get excited about the use cases (and you should be excited), the conversation you need to have internally is about data hygiene. Get that right, and the Claude + Salesforce connection becomes genuinely powerful.
The Efficiency Problem: Agentforce Isn’t the Issue. Your Data Might Be
This one I want to speak to directly, because I heard a version of it in almost every conversation I had on the floor. A lot of organizations have tried Agentforce, or are thinking about trying it, and the question underneath is always: “but does it actually work?” And honestly, from what I saw and heard at World Tour, the answer is – it depends. Not on the technology, but on what you’re pointing it at.
Here’s my take: Agentforce works. The demos are real. The customer stories are real. But when organizations don’t see the efficiency gains they expected, it’s almost never because the agent failed. It’s because the data it was given was incomplete, inconsistently structured, or just hadn’t been maintained. Garbage in, garbage out, and that’s not a new problem, but it becomes a much more visible one when an AI agent is the thing surfacing the gaps.
The same goes for the agent setup itself. An Agentforce implementation that hasn’t been properly configured, where the instructions are vague, the guardrails aren’t set, or the handoff to humans isn’t designed well, will produce results that feel frustrating and unreliable. That’s not Agentforce being bad. That’s an implementation problem.
What Salesforce was essentially saying at World Tour, between the lines, is this: the platform is ready. The question is whether your organization is. And that means looking honestly at your data quality, your process definitions, and your governance before you point an agent at your customers or your team. It’s not glamorous advice, but it’s the right one.
Salesforce Acquires Fin (Formerly Intercom) for $3.6 Billion
This was announced just days before the event and dominated a lot of the hallway conversations. Fin, which rebranded from Intercom earlier this year, is an AI customer service agent that resolves an average of 76% of support volume end-to-end, across chat, email, WhatsApp, SMS, phone, and Slack. Salesforce is paying $3.6 billion for it, with the deal expected to close in Q4 of their fiscal year 2027.
The strategic logic is clear. Agentforce has been Salesforce’s highly customizable, deeply integrated enterprise agent platform. Fin brings something different: packaged, fast-to-deploy, AI-native customer service that already has 30,000 companies using it. Together, they’re positioning for customers at every stage: from “I want an agent running this week” to “I need something fully bespoke built on my Salesforce data.” For organizations thinking about member service or customer support automation, this acquisition is worth watching closely.
Revenue Cloud: From CPQ Migration to Agentforce Revenue Management, and What We’re Building on Top of It
Revenue Cloud didn’t get a big headline moment on the main stage, but the breakout sessions told a more interesting story. The standout was a live customer Q&A where a large fintech shared how they moved off legacy CPQ systems and onto Revenue Cloud with Agentforce Revenue Management. What struck me was how honest it was not “Salesforce is better,” but a real conversation about what stalled, what changed for their sales teams day-to-day, and what governance looked like for getting an agentic revenue process into production.
The broader narrative Salesforce is pushing for Revenue Cloud is AI agents sitting inside the quote-to-cash process, automating subscription changes, flagging renewal risks, handling routine billing queries. For businesses managing renewals, tiered pricing, and self-service subscriptions, this is the direction of travel.
And it’s one we’ve been building toward at Advanced Communities. Our Revenue Cloud Quick Start includes exactly the infrastructure that makes this possible in practice. That includes a customer self-service subscription portal built on Experience Cloud, no extra Salesforce licenses (no custom development required) where members and customers can manage their own subscriptions, update payment details, and handle renewals without calling anyone. We’ve also built a WhatsApp and SMS connector that links directly into Salesforce, so renewal reminders, payment confirmations, and membership updates reach people on the channels they actually use, not just in a portal they have to remember to log into. Both are part of the Quick Start package, which covers up to 20 products and 5 bundles in a fixed-price, 6 to 8 week engagement. For a deeper look at the architecture behind this, read our guide to Revenue Cloud ordering through chat and WhatsApp – Salesforce Headless 360: How to Enable Revenue Cloud Ordering Through Chat and WhatsApp.
Learn more about Revenue Cloud Quick Start with the WhatsApp and SMS connector included for free.
Experience Cloud: Headless Is the New Normal
There wasn’t a dedicated Experience Cloud session that stopped me in my tracks, but the Headless 360 announcement and the broader Agentic Enterprise framing are essentially an Experience Cloud story repackaged for 2026. The message is that the old model with Salesforce-hosted pages, standard templates, out-of-the-box components is giving way to an API-first, agent-accessible layer where your digital experience can live anywhere, on any surface, and still run on Salesforce data and logic underneath.
For us at Advanced Communities, this matters in a specific way. Our clients use Experience Cloud-based portals as their primary digital touchpoint: for member management, event registration, self-service, and more. The shift toward headless isn’t a threat. It’s an opportunity to bring agent-powered interactions directly into those portals (member service, renewal nudges, event recommendations) without rebuilding from scratch. I think the next 18 months will see Experience Cloud partners moving fast here, and we’re already thinking about what that looks like for our products.
What This Means If You’re Already on Salesforce CRM
If your organization is already running on Salesforce CRM, the honest message from World Tour is that you’re sitting on more than you’re probably using. Most Salesforce customers I talk to are using it as a very expensive database: contacts in, reports out, and a lot of manual work in between. What the Agentic Enterprise direction means in practice is that the platform you’re already paying for is becoming capable of doing a significant chunk of that manual work for you. The data is already there. The customer relationships are already there. The processes are already mapped, at least loosely, inside your org. What’s changing is that you can now put an agent on top of all of that, one that doesn’t just surface information but actually takes action. Chases a renewal. Responds to a member query. Updates a deal stage. Flags an account that’s gone quiet. The shift from “Salesforce as a system of record” to “Salesforce as a system that acts” is real, and it’s happening now, not in two years. For organizations already on the platform, this isn’t a reason to go shopping for something new. It’s a reason to go back and look hard at what you have: how clean your data is, how well your processes are defined, and where the biggest manual bottlenecks are today. Because those bottlenecks are exactly what Agentforce is designed to remove. The organizations that will move fastest aren’t the ones starting from scratch. They’re the ones who already have Salesforce, already have the data, and are ready to stop treating it like a filing cabinet.
Final Thought
World Tour London always moves fast, and there’s always more happening in the breakouts than you can possibly catch. But the overall feeling I left with is that this year felt different from previous editions. The conversations at the booths weren’t about whether to do AI — they were about how to get out of pilot mode and into production. The announcements backed that up: Headless 360, Agentforce Coworker, Agentforce Revenue Management in customer stories, Fin coming into the portfolio. It all points in the same direction.
For us at Advanced Communities, it confirms that the work we’re doing on Revenue Cloud, Experience Cloud, and Agentforce-powered member experiences is on the right side of where the platform is going. The infrastructure is maturing. The question now is execution.
If you want to talk through what any of this means for your organization, get in touch.
By Anna Babur, Head of Marketing at Advanced Communities