AI+ Studio Deployment for Multi-Screen ACWs
Updated
You can now deploy AI Autofill for specific ACW screens in AI+ Studio. AI+ Studio lets you target specific screens within an ACW (After Call Work) when setting up AI Autofill deployments. When creating or editing an AI Autofill deployment, you can now filter by *ACW Name* to select which workflow (ACW) you want your deployment to apply to.
Previously, Autofill worked only for a single screen per ACW. If your ACW had multiple screens (for example, “Customer Details,” “Issue Summary,” “Resolution Summary”), the same AI prompt would try to run across all of them — often leading to incomplete or incorrect results.
With the new ACW Screen filters, you can now:
Create separate Autofill deployments for each screen within an ACW.
Ensure the right AI logic runs in the right place.
Improve accuracy, control, and efficiency of your AI Autofill workflows.
Prerequisites
Make sure the following prerequisites are complete before you configure a deployment:
AI Prefill is enabled for the ACW. For more information, see AI-Powered ACW Prefill.
The ACW contains the required screens and fields.
Input fields include clear AI Prefill instructions so the model can generate accurate values.
The required Dynamic Properties (DP) are enabled in your environment. For more information, see Configuring AI-Powered ACW Prefill.
Deployment Options for Multi-Screen ACWs
There are two ways to configure AI Prefill for a multi-screen ACW:
- Use the default deployment
- Create a custom deployment
Sprinklr recommends using the default deployment first because it is designed to support most use cases with minimal setup. Custom deployment is intended only for advanced scenarios where additional prompt-level control is required.
Option 1: Use the Default Deployment (Recommended)
Sprinklr provides a default deployment for ACW Prefill. This deployment is designed to support multi-screen ACWs and single-screen ACWs.
To use the default deployment:
From the Launchpad, search for AI+ Studio and open it.
Click AI Use Cases.
Select the Sprinklr Service tab, and under Agent Assist, select ACW Prefill.
On the Deployments page, locate the Sprinklr Default Deployment.
Ensure the deployment is active. If multiple deployments exist, adjust the deployment priority as needed.
Trigger the ACW and verify that:
- screens with input fields show the AI loader and display prefilled values, and
display-only screens (Asset, CTA, Description, Knowledge Base Article) do not show the loader.
Example of multiple screens with Input Fields
Note: The default deployment cannot be edited. If it does not satisfy your business requirements, create a custom deployment.
Option 2: Create a Custom Deployment
Use a custom deployment only if the default deployment does not meet your requirements. A custom deployment gives you more control over the prompt and output structure.
Important: For multi-screen ACWs, screen-specific behavior must be handled in the prompt. Do not rely on screen-level deployment filters for this use case. Use prompt conditions based on the ACW screen name instead. This is the recommended approach for custom multi-screen configurations.
To create a custom deployment:
From the Launchpad, search for AI+ Studio and open it.
Click AI Use Cases.
Select the Sprinklr Service tab, and under Agent Assist, select ACW Prefill.
Click + Deployment to create a new deployment, or open an existing custom deployment.
Enter the deployment details, such as name, priority, and description.
Open the deployment pipeline.
Add or edit the Prompt node.
In the prompt, add screen-specific conditional logic using the ACW screen name. Each condition must return only the fields that belong to the corresponding screen.
Deploy the configuration.
Test the ACW to confirm that each screen behaves as expected.
Add Screen-Specific Logic in the Prompt
For custom deployments, define screen-level behavior inside the prompt. The prompt should check the screen name and return the correct JSON output for that screen only. This helps ensure that each screen receives only the values relevant to its input fields.
Use the following approach:
If the current screen is Screen 1, return JSON for the fields on Screen 1 only.
If the current screen is Screen 2, return JSON for the fields on Screen 2 only.
Do not return fields for screens that are not currently being processed.
Example Prompt Logic
If SCREEN_NAME equals "Customer Details"
Return output only for the fields on the Customer Details screen.
If SCREEN_NAME equals "Resolution Summary"
Return output only for the fields on the Resolution Summary screen.
The output returned by the prompt must match the ACW field structure expected by AI Prefill. For sample prompt formatting, refer to Sample Prompt for AI Autofill.
Validate the Deployment
After deployment, trigger the ACW and verify the following:
Screens with input fields show the AI Prefill loader.
Screens with only display components do not show the loader.
Input fields are prefilled correctly with AI-generated suggestions.
Each screen receives only the fields relevant to that screen.
Best Practice
Always start with the default deployment. Create a custom deployment only if the default deployment does not meet your requirements. This approach reduces setup effort and helps you benefit from ongoing improvements to the default deployment.
Key Benefits
Higher accuracy: Prompts and logic match each screen’s context.
Better control: Manage multiple screens under the same ACW easily.
Screen-specific targeting: Apply AI Autofill only where it’s needed.
Optimized performance: Avoid unnecessary AI usage and improve efficiency.
Related Articles