Create New Model

Configure AI Model(s) in Predict4Dynamics

Go to Predict AI Models in the Predict4Dynamics app. Either create a new model record or select ready-to-use sample models.

1. Model Name

Enter a unique Model Name for identification. This name helps distinguish different predictive models you create for various CRM scenarios.

2. Select a Table to Predict

From the Select a Table to Predict dropdown, choose the Dynamics 365 table (entity) you want the model to run predictions on.

• Supports both OOB (out-of-box) and Custom entities

• Once configured, predictions will automatically appear in the Predict4Dynamics side panel for each respective record.

This step ensures the prediction engine understands which dataset to analyze.

3. Select Existing Data to Train Your Model On

In this field, enter a FetchXML query that retrieves the historical data needed for model training.

• The FetchXML should match the entity chosen in the previous step.

• The system uses this dataset to learn from past records and generate more accurate predictions.

Optimizing your FetchXML ensures cleaner data and better model accuracy.

4. Select Columns That Influence the Prediction

Choose the fields (columns) from the selected entity that directly impact the prediction outcome.

• These selected fields become the input features for model training.

• Predict4Dynamics uses these column values from the FetchXML results to understand patterns and behaviors.

Selecting high-quality influencing fields significantly improves prediction performance.

5. Select a Column to Predict

Pick the target column whose value you want the model to forecast.

• Example (Lead entity):

• Status → Predict whether a Lead will be Qualified or Disqualified

• Rating → Predict whether a Lead is Hot, Warm, or Cold

This helps automate decision-making by predicting real-time outcomes for CRM records.

6. Prediction Settings

Prediction Trigger On

By default, Create and Update are selected.

• On Create: Prediction runs when a new record is added.

• On Update: Prediction runs when a selected influencing column is modified.

Users can customize triggers based on business needs—only Create, only Update, or both.

This setting ensures predictions are always generated at the right moment.

7. Prediction Explainability

Enable or disable Explainable AI (XAI) insights.

• Yes: The system provides a human-readable explanation of why the prediction was generated.

• No: Only the predicted value is shown, without explanation.

This feature increases transparency and makes predictions easier to trust and interpret.

8. Display Prediction on Form

Select the CRM Form where you want the prediction result to appear. This ensures users can see predictions directly on the record form without navigating elsewhere, improving workflow efficiency.

9. OpenAI Settings

By default:

• Model Name: gpt-model

• API Version: 2025-01-01-preview

If your Azure portal uses a different deployed model:

• Set Yes in OpenAI Configuration Settings

• Enter your custom Azure OpenAI model name and API version

This allows Predict4Dynamics to integrate seamlessly with your existing Azure OpenAI setup.

For any queries, reach out to us at [email protected]