AI Sentiment in AI+ Studio lets you customize message-level sentiment analysis to fit your business needs. You can refine how sentiment is interpreted by editing prompts and validating results through bulk testing. This builds on the existing message-level AI Sentiment capability available in the platform.

Sentiment analysis identifies the emotional tone of text, such as customer messages, reviews, or conversations. By customizing prompts, you can adapt sentiment detection for your industry, terminology, and use cases.

Prerequisites

​

To configure and manage AI Sentiment deployments, ensure that you have AI+ Studio enabled in your environment and the following permissions under AI+ Studio → Sentiment Enrichment:

• View: Access and review sentiment configurations

• Edit: Modify deployments and prompt configurations

• Delete: Remove existing deployments

Create and Configure an AI Sentiment Deployment

Follow these steps to configure an AI Sentiment use case:

• Open AI+ Studio from the Sprinklr Launchpad.

• Select Deploy Your Use Case.

• Under Platform use cases, select AI Sentiment.
  You are redirected to the Deployments record manager.

• Click + Add Deployment in the top-right corner.

• In the Basic Details screen, enter the required details:

Basic details fields

Click Save.
You are redirected to the Pipelines screen.

Customize the Prompt Node

In the pipeline, configure the Prompt node to define how sentiment should be evaluated.

• Update the prompt based on your business context
• Include examples or instructions to improve accuracy

For more details on configuring the Prompt node, see Configure Prompt Node in AI Deployments.

After updating the prompt, save the Pipeline as a draft.

Impact of Prompt Customization

Customizing prompts for message-level AI Sentiment affects multiple downstream modules and metrics.

Important: Any change to sentiment logic can significantly impact reporting, automation, and agent evaluation across the platform.

Key Impact Areas

• CSAT (Customer Satisfaction Score)

  • CSAT calculation depends on message-level sentiment signals. 

• Case-Level Sentiment

  • Case sentiment is derived by aggregating message-level sentiment across the conversation. 

• Agent Quality (QM) Scores

  • Empathy Score → triggered by customer sentiment
  • Effectiveness Score → evaluates responses based on sentiment
  • Accountability Score → based on bot/auto-reply sentiment
  • Promptness Score → uses the latest detected sentiment

• Smart Alerts (Listening Module)

  • Alerts based on sentiment thresholds may trigger differently.

• AI-Powered Case Routing

  • Cases are routed based on sentiment and priority, so routing outcomes may change.

• Contact Driver Analysis

  • Aggregated sentiment metrics on dashboards will shift.

• Reporting and Analytics

  • Metrics impacted include:

    • Sentiment volume (Positive/Neutral/Negative)
    • Agent performance breakdown
    • Copilot summaries
• Advocacy and Engagement Dashboards

• Inbound sentiment trends and visualizations will change.

​

Recommendation:

Always validate prompt updates in a test deployment and review downstream impacts before enabling in production.

Bulk Test the Deployment

Bulk testing lets you validate your sentiment configuration across multiple inputs.

Open Bulk Test

1. In the Deployments record manager, locate your deployment.
2. Click the vertical ellipsis (⋮) next to it.
3. Select Bulk Test.

A third pane opens showing previous test runs.

Run a New Bulk Test

To create a new bulk test:

1. Click the + (plus) icon in the top-right corner of the pane.

2. In the Initiate Bulk Test screen, configure the following:

   • Name: Enter a unique test name
   • Bulk Execution Threshold: Define the maximum number of requests per run

3. Upload an input file in .xls, .xlsx, or .csv format.

   • Each column represents one input variable
   • You can download a template file, update it with test data, and upload it
4. Click Save.

The test appears in the pane with a Queued status. Once execution completes, the status updates to Completed or Failed.

View Bulk Test Results

1. Click the eye icon next to a test run.
2. You are redirected to the Bulk Test Detailed Log screen.

Use this view to analyze results and refine your prompt.

You can create a customized AI Sentiment deployment, test it at scale, and refine sentiment interpretation based on your use case.

Use BYOK or BYOM for Message-Level AI Sentiment

You can use Bring Your Own Key (BYOK) or Bring Your Own Model (BYOM) for message-level AI Sentiment through AI+ Studio.

• BYOM lets you integrate your own LLMs and control model behavior and hosting. 
• BYOK enables you to use your own provider credentials for supported models.

For setup details, refer to BYOM in AI+ Studio 

Prerequisites for BYOK/BYOM Deployments

Before promoting your deployment to production:

• Validate your configuration on a smaller dataset using a custom deployment in AI+ Studio.
• Use bulk testing to ensure accuracy and stability before scaling. 

Latency Requirement for Account-Based Messages

For account-based messaging workflows:

• The system enforces a 2-second response latency.

• If your model or external service exceeds this limit:

  • The message is automatically classified as Neutral (default fallback).

• You should:

  • Test latency consistency during validation.

• Ensure the deployment meets SLA requirements before production rollout.

Next Steps

• Refine your prompt using insights from bulk testing
• Increase test coverage with diverse message samples
• Deploy the configuration to production once validated