AI in CX Enterprise
This guide covers AI capabilities in Adobe CX Enterprise: generative AI, AI Assistant, Agent Orchestrator, Experience Platform Agents, CX Enterprise Coworker, and MCP.
AI capabilities overview
Start here for a primer on where and how AI is used across CX Enterprise:
- About generative AI describes which CX Enterprise applications support generative AI and AI Assistant, and how they compare.
- About agentic AI explains how Experience Platform Agents work in both existing CX Enterprise applications and AI-first applications, and lists the agents available in each.
- AI monitoring covers the dashboards that track agent adoption, usage, feedback, and AI Credit consumption.
- AI credits consumption explains how agent jobs consume AI Credits, with estimated consumption rates by agent and job type.
- CX Enterprise agentic tools cover additional agentic skills and tooling that extend CX Enterprise agents (video tutorials).
AI Assistant
AI Assistant is a conversational, generative AI tool available in Adobe Experience Platform-based applications. Use it to gain product knowledge, troubleshoot problems, find operational insights, and access Experience Platform Agents, all through natural language prompts in a full-screen or rail view interface.
To learn how to navigate the interface, read the AI Assistant UI guide. To see example prompts by agent, see the prompt library.
Agent Orchestrator and Experience Platform agents
Agent Orchestrator is the agentic layer that powers Experience Platform Agents. When you ask AI Assistant a question, Agent Orchestrator plans the work, calls on the specialized agents needed to answer it, and returns a unified response, all with human oversight.
The following Experience Platform Agents are documented in this guide:
For the full list of agents, the applications each supports, and eligibility requirements, see Agentic AI in CX Enterprise.
Coworker
Coworker is an agent-first evolution of AI Assistant that automates customer experience and marketing workflows, so your team can focus on business goals instead of routine execution. Instead of asking one question at a time, you describe a goal. Coworker plans, executes, validates, and returns the finished work for your approval. Coworker includes:
- Coworker Chat: A conversational interface for exploring your data, validating audiences and journeys, and completing multi-step tasks across CX Enterprise applications.
- Coworker Campaigns: An AI-native application that consolidates campaign briefing, audience building, content generation, journey design, and proofing into a single conversational experience. It uses built-in templates, best practices, and prompting guidance to help small, agile teams launch campaigns quickly.
- Coworker Projects (coming soon): A unified workspace for automating end-to-end customer experience orchestration workflows, helping teams coordinate tasks, approvals, and execution to drive outcomes from strategy through delivery. Documentation for Projects is coming soon.
Eligible customers are gradually being transitioned from AI Assistant and Experience Platform Agents to Coworker Chat. Read Coworker Trial to learn about trial eligibility, AI Credit usage, and how to get access.
To see Coworker Chat in action, walk through Coworker Chat in Playground, or read real-world use cases such as Validate AA to CJA migration data and Analyze CJA data.
For full product documentation on Coworker Chat, Campaigns, and Projects, see Coworker. For sandbox-to-sandbox object replication, see Sandbox Tooling Agentic Skills.
MCP
Adobe CX Coworker Gateway is the unified Model Context Protocol (MCP) endpoint for CX Enterprise. It gives MCP-compatible clients, such as Claude, ChatGPT, and Cursor, a single governed connection to the product tools your organization is entitled to use. These tools include Real-Time CDP, Experience Platform, Journey Optimizer, Customer Journey Analytics, Adobe Analytics, and Workfront.
New to CX Coworker Gateway? See Access CX Coworker Gateway tools and Install CX Coworker Gateway to get connected.
Get started
Access requirements
Your Adobe Admin must grant the appropriate permissions before you can use AI Assistant and Experience Platform Agents. Requirements vary by application; see Access in the Agent Orchestrator guide for details.
Privacy and security
AI Assistant and Experience Platform Agents prioritize privacy, security, and governance, including sandbox-specific data isolation and your existing access control policies. For full details, read Privacy, security, and governance in AI Assistant.
Best practices
To get the most value from your AI Assistant or Coworker experience, follow these best practices:
- Be specific in your prompts to obtain targeted and relevant insights.
- Verify responses by reviewing the source citations and reasoning explanations provided.
- Use context setting to make sure the most relevant data sources are used for your questions.
- Provide feedback to help improve performance and accuracy over time.
- Combine insights from multiple agents for a more comprehensive analysis.
Legal considerations
AI Assistant currently supports responses in English only, and language models occasionally make mistakes. Always verify the information provided, and use the reasoning steps included in each response to understand how it was generated. For full details, read the legal disclaimer.