Adobe retired its Experience Cloud brand at Summit 2026 in Las Vegas on April 20, replacing it with CX Enterprise, an agent-native platform that reorganizes the company’s entire marketing and customer experience stack around persistent AI agents. The announcement, confirmed by Adobe’s newsroom and detailed by MarTech, marks the most significant architectural shift in Adobe’s enterprise software since Experience Cloud launched.
Three Pillars, One Agent Layer
CX Enterprise is organized around three pillars: Brand Visibility, Customer Engagement, and Content Supply Chain. Underpinning all three is a new Adobe AI Platform with two intelligence systems. Adobe Brand Intelligence governs brand consistency across AI-powered channels, while the CX Engagement Intelligence System handles optimization across audiences, channels, and customer journeys.
More than 10 purpose-built AI agents previewed at Summit 2025 are now in production, covering site optimization, data insights, audience creation, journey orchestration, experimentation, and content optimization. According to MarTech, 1,770-plus customers are already entitled to use them through a credit-based pricing model.
Coworkers: Persistent Agents With Enterprise Memory
The centerpiece of the announcement is a new capability tier called “Coworkers,” which Adobe describes as persistent, self-learning agents with enterprise memory that can orchestrate multiple Adobe and third-party agents toward business goals. Unlike single-task agents that fire and forget, Coworkers run continuously, learn from outcomes, and can be triggered by signals or schedules.
Adobe positions Coworkers for all CX personas: developers, marketing ops, and marketers. The architecture is explicitly not a single-LLM play. Coworkers orchestrate across Adobe’s own agents and external ones, functioning as a goal-oriented coordination layer.
Governance: Two Oversight Models
Adobe disclosed two levels of human oversight. Human-in-the-Loop applies to design-time activities like campaign planning, requiring active review and approval before the agent proceeds. Human-on-the-Loop applies to consumer-facing agents like Brand Concierge, where the agent operates autonomously within preset guardrails while humans monitor and can intervene.
The company says it uses LLM-as-a-judge internally, deploying AI models to validate the outputs of other AI models before they reach end users.
Interoperability Play
The platform supports Model Context Protocol (MCP) across its products and offers reference architectures for Microsoft Copilot, ChatGPT Enterprise, Claude Cowork, and Gemini Enterprise. Adobe CX Skills, pre-built capabilities from across the platform, are available inside those partner AI environments. Three deployment models are supported: full-stack Adobe, Adobe-orchestrated within partner UIs, or bring-your-own orchestration.
Adobe also expanded Workfront into an “Agency System of Record” connecting brands directly with agency partners. WPP, Publicis, Omnicom, Dentsu, and Havas are integrated at launch.
The Competitive Context
The rebrand lands days after Salesforce announced Headless 360 at TrailblazerDX, exposing its entire platform as APIs, MCP tools, and CLI commands for agent access. Adobe’s own AI & Digital Trends Study from March 2026 found 75% of organizations cite data integration and quality as their top AI implementation challenge, 71% cite talent gaps, and 68% cite unclear ROI.
Adobe is betting that persistent agents with enterprise memory will differentiate it from the task-completion agents competitors have shipped so far. Whether enterprise buyers want agents that “learn from outcomes” and “run continuously” on their customer data is the open question Summit 2026 will likely surface.