Agentforce launched in October 2024. In less than two years, Salesforce has released four major platform versions, rebuilt its entire marketing stack on a single data foundation, and mobilized its organization with embedded engineers and dedicated AI solutioning teams for agentic deployment. They've declared, in their own words, the arrival of the Agentic Enterprise.
Salesforce Connections arrives in Chicago in early June, and this year, the platform is ready to back the vision it’s been building toward. The timing isn’t coincidental. Something is fundamentally shifting, and it’s worth understanding what before you walk through those doors.
What the Agentic Enterprise Actually Means
An “Agentic Enterprise” sounds like the next buzzword. Here’s the vision Salesforce is engineering toward:
Agent-to-agent paradigms will fundamentally reshape marketing workflows from the ground up, with an agentic fabric woven across everything.
This is the shift from a single agent handling a single use case to a system of agents that shares context, hands off work, and operates continuously across the customer lifecycle.
While this may seem ambitious, it represents a logical progression as the industry pivots from experimental AI toward operational execution at scale. The planned infrastructure to support scaled AI now exists natively: the data foundation, the orchestration layer, the reasoning engine, and the embedded analytics capability. Salesforce’s own research shows 75% of marketers are experimenting with AI, while only 32% are satisfied with how they use customer data to create relevant experiences. That gap is the core problem the Agentic Enterprise is designed to solve, a model necessary to bridge experimentation and satisfaction.
What Salesforce Has Built, and How They’re Backing the Vision
The clearest signal that Salesforce is serious isn't the volume of announcements. It's what they're doing operationally to back the vision.
Salesforce has committed to building a team of 1,000 Forward Deployed Engineers: not account managers, not solution engineers, but technical builders who embed directly with clients in three-month pods to design, deploy, and prove out agents in production. A parallel FDE Partner Network was launched, comprised of partner firms with seasoned Agentforce implementation experience.
Meanwhile, Salesforce restructured its commercial model around Agentic Enterprise License Agreements (flat-fee, outcome-linked contracts). Sixteen closed last quarter. Around one hundred are in the pipeline. Simultaneously restructuring the delivery model, the partner ecosystem, and its commercial construct is not a series of product launches; it's an enterprise operating model shift.
Salesforce’s product evolution tells the same story. The seven capabilities below are worth understanding together as an intentional architecture, each layer designed to make the others more powerful:
- Data 360: Formerly Data Cloud, now repositioned as the system of context for every agent. Neither a CDP nor a Warehouse replacement, it’s the connective tissue that every activation, every agent, and every decisioning moment runs on.
- Marketing Cloud Next: Nine acquired products collapsed onto a single Data 360-native foundation. It unifies Marketing Cloud Engagement, Personalization, Loyalty Management, and Business Intelligence. The stack is now well integrated and meaningfully expanded.
- The Analytics Layer: The future of Business Intelligence with Marketing Intelligence and Tableau Next sit on top of the same Data 360 foundation that agents run on. For the first time, the analytical surface marketers use to read performance and the activation layer agents use to make decisions draw from the same source: a shared context layer that makes insight and action speak the same language.
- MuleSoft Agent Fabric: As enterprises accumulate agents across vendors and clouds, the challenge shifts from building agents to managing them as a system. Agent Fabric provides the registry, routing, governance, and observability to do that.
- Agentforce 360: The unifying platform frame that ties agent runtime, reasoning via the Atlas Reasoning Engine, voice, and builder into one coherent development and deployment environment. Model-agnostic by design, OpenAI, Google Gemini, Anthropic, or Salesforce’s own Einstein models can all run inside it.
- Flex Credits: Salesforce recently consolidated what were formerly separate Agentforce credits and Data Cloud consumption credits into a single unified Flex Credit system. For customers, that’s one currency for agents, data, and intelligence workloads, which simplifies forecasting, budgeting, and usage across the platform.
- Headless 360: Announced April 2026, the entire platform is now exposed as APIs, MCP tools, and CLI commands, which means agents no longer need a browser or UI to operate Salesforce. The interface is no longer the platform. The platform is the infrastructure
What We’re Seeing on the Ground
Merkle is actively implementing Agentforce and Marketing Cloud Next with clients today, and what we’re learning is shaping how we advise the broader market.
Salesforce Personalization is the most mature SKU in the Marketing Cloud Next portfolio in 2026: Data Cloud-native, faster to execute, and meaningfully evolved beyond its predecessor with agent actions and Einstein Studio model interoperability. We’re evaluating it with multiple clients while also exploring the convergence capability to run it alongside Marketing Cloud Personalization (MCP), allowing for better migration experience.
The new Marketing Intelligence, built on core and backed by Data 360, is a clean break from the Datorama MCI stack, putting reporting where the data already is which reduces unnecessary movement.
Marketing Cloud Advanced is showing up in many Marketing Cloud Engagement discussions about what the future looks like and how to take advantage of the MC Next+ coexistence capabilities. Marketing Cloud Advanced may not be fully ready as a standalone for large, multi-dimensional enterprise brands, but we are implementing Unified Profiles and Data 360 segmentation using Flow-based journeys which simplify the overall marketing architecture and enable agentic capabilities to tie in.
On the Agentforce side, the use cases finding the most durable footing are those with well-defined inputs and measurable outputs, such as service deflection, case resolution, and sales pipeline management. Capabilities like Einstein-powered segmentation and data discovery show real promise but still require human oversight to perform reliably end-to-end. The pattern we’re seeing consistently: the tighter the workflow, the faster the value.
What We’re Watching For
Connections 2026 is where Salesforce moves from architectural vision to operational guidance, and where the marketing and commerce community gets its first real look at the Agentic Enterprise in practice. Where we’ll be keenly focused: migration clarity for Marketing Cloud Engagement customers.
The stack has consolidated in theory; Connections is where the “how and when” has to get real. Production proof points for Personalization Decisioning, Segment Intelligence, and Paid Media Optimization must move beyond the demo environment. Brands want to understand the requisites for scaling agents and data readiness to truly operationalize this ecosystem, specifically, what it takes to get a Data 360 environment clean and connected enough for agents to run reliably on it. We’re hoping Chicago delivers practical answers and enablement stories, not just demos.
By the end of 2026, the distance between organizations running governed multi-agent systems on a unified data foundation and those still cycling through isolated pilots will become measurable: in revenue, in retention, and in the speed at which they can respond to customers. The architecture is ready, the separation is beginning.
