Adobe has decided to take another step forward in the race to bring agentic artificial intelligence to the heart of enterprise marketing. At Adobe Summit 2026, the company introduced
Adobe CX Enterprise, a new end-to-end system that aims to unify AI agents, reusable skills, connection points based on Model Context Protocol (MCP), and an intelligence and governance layer to automate customer experience flows more reliably, auditable, and personalized. The proposal directly addresses one of the industry’s major debates: how to move from isolated AI pilots to truly operational processes within large organizations.
The move is significant. Adobe states that more than 20,000 global brands already build part of their business on its technology and that Adobe Experience Platform (AEP), the contextual foundation supporting this new offering, already powers over one trillion experiences annually. With this starting point, CX Enterprise is not presented as a standalone tool but as a new orchestration layer designed to coordinate data, content, decisions, and execution at a time when AI agents are beginning to integrate into daily marketing, loyalty, and personalization workflows.
The architecture announced by Adobe revolves around four main components, according to the company: agents integrated into its applications, a catalog of reusable skills, developer tools, and a new “co-worker” designed to coordinate complex tasks among multiple agents. All of this is supported by Brand Intelligence and Engagement Intelligence, two engines that Adobe places at the center of its new vision for customer experience in the agentic era.
| Component | What it offers |
|---|---|
| Agents in Adobe Applications | Automate engagement tasks, content supply chain, and brand visibility |
| Skills Catalog | Allows packaging reusable instructions to create custom workflows |
| Developer Tools | Provide access to skills, MCP servers, and infrastructure for tailored use cases |
| CX Enterprise Coworker | Coordinates multiple agents and tools to execute business objectives with human oversight |
From experimental AI to real orchestration
The key to the announcement lies less in the product name than in the problem it seeks to solve. Over the past two years, many companies have added assistants, copilots, or partial automations, but few have succeeded in integrating that AI layer into complete marketing and customer experience workflows. Adobe aims to fill that gap. Its message is that the next phase isn’t about having more models, but about orchestrating actions, data, and content more effectively through business rules, human control, and traceability.
To support this promise, the company relies on Adobe Experience Platform Agent Orchestrator, the foundation it has been building since 2025 to implement agents in its ecosystem. Adobe explains that this orchestrator combines a conversational interface, a reasoning engine, a knowledge base, and composition and configuration tools so agents can perform complex tasks with context, validation, and repeatable outcomes. In other words, Adobe is attempting to turn agentic AI into an enterprise infrastructure component, not just an interaction layer.
This ambition is also reflected in how it defines its two new intelligence engines. Adobe Brand Intelligence is envisioned as a continuous learning system capable of capturing shifting brand signals, while Adobe Engagement Intelligence is presented as a decision engine optimized for customer lifetime value. The idea is significant because it shifts the focus from mere content generation to something more complex: ensuring each automated action aligns with brand identity and measurable business objectives.
An open ecosystem, but with Adobe’s control
One of the most interesting aspects of the announcement is that Adobe does not want to confine this offering within its own ecosystem. The company emphasizes that CX Enterprise is built with a composable, open architecture compatible with tools from AWS, Anthropic, Google Cloud, IBM, Microsoft, NVIDIA, and OpenAI. This interoperability is not just lip service: Adobe has confirmed that its Marketing Agent is now generally available on Microsoft 365 Copilot and in beta on platforms like Amazon Quick, Anthropic Claude Enterprise, ChatGPT Enterprise, Gemini Enterprise, and IBM watsonx Orchestrate.
This approach reflects a reality in enterprise: daily work rarely resides within a single platform. Marketing, operations, sales, and creative teams navigate across different suites, corporate systems, CRMs, collaboration tools, and third-party applications. Adobe seems to recognize that if its agents are to become a useful layer, they need to operate within these environments without requiring complete migrations. It’s a pragmatic, likely necessary strategy in a market where interoperability now weighs as much as model capabilities.
At the same time, Adobe aims to maintain a role as an arbiter. The company emphasizes that it retains a layer of quality, governance, and control over outcomes, highlighting in its Agent Orchestrator documentation its standards for security, privacy, and ethical review of AI. This blend of openness and control is perhaps the most delicate part of the proposal: Adobe wants to be everywhere, but not at the expense of losing its role in validating, contextualizing, and governing the execution.
Coworker, the closest piece to the “executive agent” concept
Among the components announced, Adobe CX Enterprise Coworker likely carries the most strategic weight. Adobe describes it as a coordinator capable of translating a business goal into a sequence of actions involving multiple agents. The company’s illustrative example is compelling: if a team wants to boost cross-sell performance by 3%, Coworker could gather audience segments, creative assets, performance data, and insights to design a plan, obtain human approval, execute, and monitor it. Adobe indicates this feature will be generally available in the coming months, so it should be seen as a launch with some of its value still pending full deployment.
This introduces an important nuance. Adobe has presented a powerful vision aligned with current enterprise software trends, but some parts of the offering are still in the rollout phase. The company includes forward-looking statements in its message and acknowledges that some capabilities, innovations, and benefits depend on future developments. Simply put: the strategy is clear, but full execution still has a way to go.
Nevertheless, the announcement is significant because it reveals where big marketing software is heading. Adobe isn’t just selling more automation or a new chatbot for business teams; it’s trying to redefine the operational layer of customer experience through agents, skills, data context, and governance rules. If successful, the conversation will shift from “what model does a company use” to “what system truly coordinates data, content, and decisions at scale.” And Adobe aims to lead that shift ahead of many competitors.
Frequently Asked Questions
What exactly is Adobe CX Enterprise?
It’s an end-to-end agentic system introduced by Adobe at Summit 2026 to manage the customer lifecycle using AI agents, reusable skills, MCP points, and an intelligence and governance layer.
What differentiates CX Enterprise from other marketing AI assistants?
Adobe positions it not as a standalone assistant, but as an orchestration platform combining AEP data, coordinated agents, auditable decisions, and human oversight to execute complete customer experience flows.
Which external platforms will Adobe CX Enterprise be compatible with?
Adobe announced interoperability with AWS, Anthropic, Google Cloud, IBM, Microsoft, NVIDIA, and OpenAI. Additionally, its Marketing Agent is available on Microsoft 365 Copilot and in beta on platforms like ChatGPT Enterprise, Claude Enterprise, Gemini Enterprise, Amazon Quick, and IBM watsonx Orchestrate.
When will Adobe CX Enterprise Coworker be available?
Adobe has indicated that CX Enterprise Coworker will become generally available “in the coming months,” so it does not yet have universal availability.