Adobe CX Enterprise Coworker modules
Use the Adobe CX Enterprise Coworker connector in Adobe Workfront Fusion to send a natural-language prompt to Coworker and use its response in a scenario. Coworker interprets your request and can use tools available to your account to carry out the work.
The connector provides two modules:
- Run a Prompt starts a new request and, by default, waits for the result.
- Get Execution Status checks a request that you started earlier.
Access requirements
| table 0-row-2 1-row-2 2-row-2 layout-auto html-authored no-header | |
|---|---|
| Adobe Workfront package |
Any Adobe Workfront Workflow package and any Adobe Workfront Automation and Integration package Workfront Ultimate Workfront Prime and Select packages, with an additional purchase of Workfront Fusion. |
| Adobe Workfront licenses |
Standard Work or higher |
| Product | If your organization has a Select or Prime Workfront package that does not include Workfront Automation and Integration, your organization must purchase Adobe Workfront Fusion. |
For more detail about the information in this table, see Access requirements in documentation.
Prerequisites
YOu must have an Adobe account that has access to Adobe CX Enterprise Coworker.
Connect Adobe CX Enterprise Coworker to Workfront Fusion
Sign in with an Adobe account that has access to Coworker and any services needed for your request.
To create a connection:
-
In the Adobe Fusion MCP module, click Add next to the Connection field.
-
Fill in the following fields:
table 0-row-2 1-row-2 2-row-2 3-row-2 4-row-2 layout-auto html-authored no-header Connection name Enter a name for this connection. Environment Select whether you are connecting to a production or non-production environment. Type Select whether you are connecting to a service account or a personal account. Client ID Enter or map an optional override for this connection. Otherwise, use this account type's platform-managed clientId. Register the displayed Fusion callback URL for the dedicated Coworker account.. Client secret Enter or map an optional override for this connection. Otherwise use this account type's platform-managed clientSecret. Use credentials issued for the configured IMS environment. -
Click Continue.
If you cannot sign in or use a required service, contact your administrator.
Adobe CX Enterprise Coworker modules and their fields
Get Execution Status
This action module checks the current state of a task started by Run a Prompt.
This module checks the task once. It does not start a new task or keep checking until the task completes. If the result is still pending, check again later.
Run a Prompt
This action module sends an instruction to Coworker, such as summarizing text or gathering information using tools available to your account.
Each run sends the prompt once and starts a new conversation. You do not need to select an AI model or provide a separate AI-provider key.
Wait for completion
When Wait for completion is on, the module checks the task after a five-second wait. If the task is still pending, the module waits another five seconds before checking again. Each check also takes time to receive a response. Checking the status does not send the prompt again.
Waiting stops when the task completes, fails, is canceled or rejected, or needs your input or authorization. If Coworker provides an immediate answer, the module returns it without further checks.
When Wait for completion is off, the module returns the first result without waiting for the task to finish. Inspect State, Completed, and Requires attention to decide what to do next. A task can already be completed when the first result arrives.
Waiting is subject to Fusion’s module timeout. If the module times out or stops, Coworker might continue working. In that case, you might not receive an output or a task ID.
For longer work, turn waiting off and save the initial Task ID so you can check the task later. Do not automatically run the prompt again after a timeout. The first request might already be in progress, and repeating it can duplicate changes.
Module output
Both modules return the following information when they finish without an error.
Use Final assistant text for the answer instead of combining all text artifacts. Artifacts can include progress descriptions or other intermediate text.
Requests that need attention
Coworker might need more information or authorization before it can continue. When this happens, Requires attention is true and Completed is false. Route the result to a person instead of treating it as completed work.
These modules cannot answer a follow-up question or approve an action on your behalf. Get Execution Status can check whether the task’s state has changed, but it does not resolve the request for input or authorization.
Use previous answers
Every Run a Prompt execution starts a new conversation. To refer to an earlier answer, include the relevant question and answer as text in your new prompt. This gives Coworker context, but it does not resume the previous conversation.
Context ID identifies the conversation associated with the result. You cannot supply it to these modules to continue a conversation.
Task states
A pending result, or a result that requires attention, is not a successful completion, even if the module returns without an error.
Examples
Summarize an update
-
Add Run a Prompt and select your connection.
-
Enter the following in the User prompt field:
code language-text Summarize the following project update in three bullet points. Use only the supplied text; do not call tools or change any records. The planning workshop is complete. Two dependencies need confirmation. The team will review the delivery estimate on Friday. -
Leave Wait for completion on and Include raw result off.
-
If Requires attention is true, route the result for human review. If Completed is true, map Final assistant text into your next step.
AI-generated wording can vary. Review results where accuracy matters.
Check a longer task later
-
Add Run a Prompt, select your connection, and enter your instruction.
-
Turn Wait for completion off.
-
Inspect the initial result.
- If the task is completed, process the result.
- If the task requires attention, route the result for review.
- If the task is pending, save the Task ID as text and note which connection you used.
-
Later, use Get Execution Status with that connection and Task ID.
-
If the task is still pending, arrange another check within your scenario’s limits. Process the result once on completion, and handle attention or errors separately.
Do not run the original prompt again just to check its progress.
Usage considerations
- Coworker can use tools and perform actions allowed by your account, including changes to records. Use trusted prompts and an account with appropriate access.
- These modules do not provide a way to restrict individual tools. Instructions such as “do not change anything” describe your intent, but they are not an enforced read-only mode.
- Available tools and skills vary by account. The connector does not provide a skill selector or guarantee a particular action, such as creating a journey or updating a project. Do not assume that requests that need approval will complete unattended.
- The connector starts and checks Coworker requests. It does not monitor project changes or automatically save answers to another application. Add the appropriate trigger, storage, and follow-on modules to your scenario as needed.
- AI-generated answers can be incomplete or inaccurate. Validate important results before you use them to make changes or decisions.
- The connector does not automatically retry a failed request. Make sure your scenario’s error handling does not repeat a prompt whose outcome is uncertain.