Adobe brings CX Coworker to ChatGPT, Claude and Copilot
Adobe brings CX Enterprise Coworker to ChatGPT, Claude and Copilot, with new MCP connectors to ad and marketing systems.
Adobe has put its CX Enterprise Coworker marketing agent inside ChatGPT, Claude, Microsoft Copilot and other enterprise AI platforms. Its November 4 announcement also introduced prebuilt connectors that can link assistant-led work to advertising and customer-experience systems.
The update gives enterprise marketers more places to start campaign analysis, planning and orchestration. The central question is whether the same governance, data permissions and measurement standards hold when the interface sits outside Adobe.
Table of contents
Jump to each section:
- What the November rollout changes
- Where Coworker is available now
- What the new MCP connectors add
- The competing enterprise agent layers
- What marketers should test before connecting Coworker
What the November rollout changes
Adobe had already announced general availability of CX Enterprise Coworker in June. That earlier milestone established the agentic marketing system itself. The November announcement is about where enterprise employees can use it: inside leading third-party AI assistants, rather than only through an Adobe-led workflow.
The company says Coworker can draw on brand, customer and channel context while coordinating marketing work under human oversight. Anjul Bhambhri, Adobe's senior vice president of engineering for customer experience orchestration, described the problem this way: “Today, marketers often use AI tools in isolation.” The new distribution is meant to connect those isolated work surfaces to enterprise customer-experience systems.
That distinction is material for marketing operations teams that already rely on several AI assistants. Availability inside an assistant creates a new starting point for work; it does not by itself establish that every enterprise account has permission to use every campaign action.
Where Coworker is available now
According to Adobe, Coworker is now accessible within ChatGPT as a plugin, Anthropic's Claude, Databricks Genie, Gemini for business and Microsoft Copilot. These are the available integrations named in the November announcement.
Adobe separately lists Amazon Quick, IBM and Snowflake as future destinations. Their inclusion in the partner roadmap should not be mistaken for existing availability. PYMNTS' November 4 report independently corroborates Adobe's named platforms and connector list.
| AI workspace | Adobe's announced status |
|---|---|
| ChatGPT (plugin) | Available |
| Claude | Available |
| Databricks Genie | Available |
| Gemini for business | Available |
| Microsoft Copilot | Available |
| Amazon Quick, IBM, Snowflake | Expected later |
What the new MCP connectors add
The second element is connectivity to marketing systems. Adobe introduced prebuilt Model Context Protocol (MCP) connectors spanning Amazon Ads, Demandbase, a Meta campaign ads MCP server, Microsoft Work IQ, RainFocus, Snowflake Cortex AI and Databricks Genie.
For example, a marketing operations team could ask an assistant for a cross-channel performance assessment, then use approved connected systems to retrieve supporting information and coordinate work. This is an illustrative workflow, not a claim that Adobe demonstrated a specific end-to-end campaign or exposed identical actions across every connection.
MCP provides a way for AI software to interact with tools and data; Adobe also references Agent2Agent (A2A) for agent interoperability. Neither standard removes the need to map access rights, approval checkpoints and audit records. The benefit comes from how those controls are implemented, not the mere presence of a connector.
The competing enterprise agent layers
Adobe's move increases pressure on enterprise martech vendors to meet teams in the AI workspace they already use. Salesforce, Oracle and Microsoft also sell parts of the customer-data, CRM, AI and orchestration stack. Adobe's emphasis is on retaining customer-experience context while the interaction moves to another company's assistant.
Adobe also named systems integrators including Accenture, Capgemini, Deloitte Digital, IBM, Infosys and TCS, alongside agencies such as dentsu, Havas, Omnicom, Publicis, Stagwell and WPP. Those are company-reported ecosystem participants, not proof that the same configurations are deployed for every customer.
For agencies, this could shift the integration brief from creating another dashboard to defining which systems, permissions and measurement rules an assistant can safely use for a client. The practical competition is increasingly about governance and connected workflows.
What marketers should test before connecting Coworker
Procurement and marketing-operations teams should begin with the official Coworker product documentation, confirm which assistant surfaces and connectors are enabled for their account, and request a sandbox demonstration involving real permission boundaries.
Four checks matter most: which customer data are accessible from an external assistant; which actions require an explicit human approval; where event logs and rollback records live; and how failures or contradictory outputs are escalated. Teams should also distinguish read-only reporting from actions that change paid-media campaigns or customer journeys.
Adobe has made its AI orchestration layer easier to reach across platforms. The harder evaluation for marketers is whether that reach comes with sufficiently predictable controls, measurement and cost for their own stack.
Key Takeaways
- Adobe made CX Enterprise Coworker available in five named external AI platforms on November 4, 2026.
- Its new MCP connectors include Amazon Ads, Demandbase, Meta campaign tools and Microsoft Work IQ.
- Enterprise teams still need to validate permissions, approval paths and logs before using agent-led campaign actions.
