Adobe's coordination‑layer pattern
- Adobe launched CX Enterprise with agentic skills, MCP servers, and a coworker coordination layer for multi‑step business actions. (timesofindia.indiatimes.com) - The product is designed to interoperate with Anthropic, Google Cloud, Microsoft, OpenAI and other vendor stacks. (timesofindia.indiatimes.com) - The offering illustrates a coordination layer sitting above models and skills that maps business goals to orchestrated agent workflows. (timesofindia.indiatimes.com)
A new layer is emerging in enterprise artificial intelligence: software that does not just call a model, but assigns multiple agents, tools, and approvals to finish a business task. Adobe put that pattern on display on April 20 at Adobe Summit with CX Enterprise. (news.adobe.com) Adobe said CX Enterprise combines AI agents, agent skills, and Model Context Protocol endpoints with what it called an “intelligence and governance layer” for auditable workflows. The launch came at Adobe Summit in Las Vegas, which runs April 20-22, 2026, with preconference events on April 19. (news.adobe.com) (summit.adobe.com) In plain terms, Model Context Protocol is a standard for connecting an artificial intelligence assistant to outside systems where data and tools live. Anthropic introduced the protocol in November 2024, and its specification says servers expose context and capabilities to host applications over JSON-RPC 2.0. (anthropic.com) (modelcontextprotocol.io) Adobe’s product pitch is that a company can describe a goal such as changing a customer journey, generating content, or adjusting delivery, and specialized agents handle the steps together instead of one chatbot doing everything. Adobe’s business blog said those agents work across content, journeys, and experience delivery and management. (business.adobe.com 1) (business.adobe.com 2) That architecture shifts the competition upward. If models and tools are becoming interchangeable through open connectors like Model Context Protocol, vendors can try to own the coordination layer that decides which model, which tool, and which approval path gets used for each step. (modelcontextprotocol.io) (anthropic.com) Adobe is also signaling that it does not want customers locked to one model stack. Its announcement said CX Enterprise is designed to work with technology from Anthropic, Google Cloud, Microsoft, OpenAI, SAP, ServiceNow, and Workday, alongside Adobe’s own applications and data systems. (news.adobe.com) That interoperability pitch matches a broader industry move around shared plumbing. In December 2025, Anthropic said it was donating Model Context Protocol to the Agentic AI Foundation under the Linux Foundation, with support from OpenAI, Google, Microsoft, Amazon Web Services, Cloudflare, and Bloomberg. (anthropic.com) Adobe is pairing the software launch with a services ecosystem meant to push it into large companies. The company said agencies including Dentsu, Havas, Omnicom, Publicis, Stagwell, and WPP, along with integrators such as Accenture, Deloitte Digital, IBM, Infosys, PwC, and Tata Consultancy Services, are building offerings around CX Enterprise. (news.adobe.com) The immediate test is whether these systems can handle long, multi-step work without losing control, context, or audit trails. Adobe’s bet is that the winning layer in enterprise artificial intelligence may be less the model itself than the software manager sitting above it. (news.adobe.com) (business.adobe.com)
