Adobe Unveils CX Enterprise Coworker at Summit 2026: Agentic AI Gets Its Own Teammate
On April 20, 2026, at Adobe Summit 2026 in Las Vegas, Adobe introduced CX Enterprise Coworker, a fully agentic AI system designed to plan, orchestrate, and execute customer experience workflows on behalf of marketing teams. It is the flagship product inside the newly unveiled Adobe CX Enterprise platform, the rebranded successor to Adobe Experience Cloud.
Where earlier AI assistants inside Adobe Journey Optimizer or Adobe Campaign simply offered copilot-style suggestions, CX Enterprise Coworker operates with what Adobe calls “persistent agency” – it holds long-running business goals, breaks them into multi-step plans, and executes once approved.
From Goal to Multi-Channel Campaign in One Brief
Adobe’s launch demo showed the Coworker reading a simple objective, “increase cross-sell on loyalty customers by 3 percent,” and autonomously assembling an audience, pulling creative from Adobe Firefly, drafting the journey, and routing approvals. Once a human signs off, the Coworker ships the campaign and monitors lift in real time, course-correcting on cadence, channel mix, and offer variants.
According to MarTech.org, Adobe is positioning the Coworker as the operational home for more than ten specialist AI agents that graduated to general availability at Summit 2026, including agents for site optimization, audience creation, journey orchestration, experimentation, and content optimization.
Built with NVIDIA, Living Inside Microsoft Copilot and Claude
Adobe made it clear that Coworker is not a closed garden. The company announced that the product is built on NVIDIA’s Agent Toolkit and runs on NVIDIA OpenShell, a policy-governed agent runtime that can be deployed on-premises or in the cloud, a direct nod to enterprise security teams who have balked at shadow-AI risk.
More striking is how far Adobe is willing to push its agents onto other companies’ turf. The Adobe Marketing Agent is now generally available inside Microsoft 365 Copilot and is entering beta in Anthropic’s Claude Enterprise, ChatGPT Enterprise, Google Gemini Enterprise, Amazon Quick, and IBM watsonx Orchestrate. The interoperability layer is the Model Context Protocol (MCP), which Adobe has adopted as the lingua franca for plugging its agent skills into third-party orchestrators.
Constellation Research called the move a platform-neutral bet and noted that Adobe’s strategy stands in contrast to vendors trying to lock agents behind a single console.
What It Means for Campaign Classic, AJO, and Marketo
For practitioners still running Adobe Campaign Classic or Marketo, the Coworker is less a replacement and more a new control plane. In journeys, it coordinates the unified Action activity that replaced the legacy Email, Push, SMS, and In-app channel nodes in February 2026, and it can hand off to Marketo Engage for B2B nurture flows via the new AJO B2B Edition Prime tier arriving in Q2 2026.
Coworker operates on Adobe Experience Platform data and respects the same Real-Time CDP governance controls that teams already use, with no new identity graph to maintain. Dynamic query logic familiar from Adobe Campaign’s segmentation engine is exposed to the Coworker as agent skills, letting it assemble audiences without human SQL.
Brand and Engagement Intelligence: The Two Brains
Flanking the Coworker are two new intelligence services announced the same day. Adobe Brand Intelligence is a continuously learning reasoning engine that captures brand signals such as tone, visual identity, and positioning so agents never produce off-brand creative. Adobe Engagement Intelligence is a decisioning engine optimized around customer lifetime value rather than click-through, intended to push personalization beyond next-best-offer heuristics.
Together, per CMSWire, the two act as guardrails that keep the Coworker’s autonomous decisions within brand and economic constraints, though the outlet cautioned that the hard work may be the operating model, meaning rewiring marketing teams to actually trust agent output.








