Overview

Zia in Zoho CRM provides built-in AI solutions such as predictions, recommendations, etc. For organizations that require more control and want to build advanced custom ML models, Catalyst’s QuickML offers a platform to create, train, and deploy models with complete flexibility over the pipeline.

These models can then be integrated into Zoho CRM through Custom AI Studio, so that customers can see a 360 degree business insight in one platform. 

Note that model creation happens entirely within QuickML—models cannot be created directly from Zoho CRM, even after integration.

Feature Availability and Permissions

• Availability: All paid editions of Zoho CRM. 
  Each module can have up to four QuickML configurations.
• Supported Data Centers: Available across all DCs except CN.
• Permissions:

1. Admins can both manage configurations and view results.
2. Standard users have view-only access to model results.
   

Why Custom AI solutions?

Addressing advanced AI requirements

Different businesses have unique goals and operate with their own data patterns and workflows. While Zia covers standard CRM solutions, others may need domain-specific solutions or proprietary models that align with their unique data and processes. A rigid AI pipeline that customers cannot modify often fails to align with these specific requirements. This is where custom ML models become essential, allowing organizations to tailor their AI solutions to their needs and ensure the solutions truly reflect their processes.

In comes QuickML.

By integrating QuickML with Zoho CRM, businesses can bring their custom-built models directly into the CRM, making decision-making more seamless and unified, all in one place for better decisions and actions .

Use Cases

Insurance claim prediction

With QuickML, insurance companies can build fully custom ML models tailored to their own claim rules, policy structures, and historical patterns. These models can predict the expected claim payout for newly submitted claims.

Once the model is trained and deployed in QuickML, it can be integrated into Zoho CRM through Custom AI Studio. Users will see the expected claim payout displayed directly within each claim record, helping teams process claims faster and maintain consistent decision-making across workflows.

Smart recommendations

An e-commerce business that tracks customer profiles, purchase history, and frequently viewed products can use this data to build a custom recommendation model in QuickML.

After the model is trained and deployed, it is integrated back into Zoho CRM. When a user opens a customer record, the CRM displays personalized product recommendations generated by their own QuickML model.

For example, if a customer regularly orders fitness accessories and recently viewed running shoes, the model can identify this pattern. Inside CRM, it may suggest related products such as sports apparel, fitness trackers, or shoe accessories. These recommendations help sales teams engage customers more effectively, increase cross-sell opportunities, and support repeat purchases, right from their CRM workspace.

Integrating QuickML in Zoho CRM

You can access the QuickML integration in the setup page of your CRM org, simply navigate to Setup > Zia > Custom AI Studio.

When you open QuickML for the first time, a Get Started page explains what QuickML is and how it helps create tailored ML integrations inside CRM.
If your QuickML account is already linked, you can proceed to configuration. If not, you’ll be guided to create an account before integrating QuickML.

Setting up QuickML integration

Step 1: Create a configuration

1. Click New Configuration in the QuickML setup page.
2. On the Create QuickML Model Configuration pop-up, do the following:
   

1. Enter a Name for the configuration.
2. Choose the module where you want to display the model’s output (for example, Leads or Deals).
3. Select the required Organization, Project, and Model Endpoint from your QuickML account.

3. Click Next to proceed.
   
   

Note: Each model endpoint in QuickML represents a deployed ML model. You can create multiple configurations under different CRM modules (up to four per module).

Step 2: Map model inputs with CRM fields

Once the model endpoint is selected, on the next screen do the following:

1. Map the input parameters to the corresponding CRM fields.
   At least one field must be mapped for the configuration to be saved.
   
2. You can also choose whether the model output should apply to:

1. All records in the module, or
2. Records matching specific conditions using filters.

3. Click Save to create the configuration.
   
   

The model is now integrated, and outputs will appear as widgets within the selected module’s record detail view.

Viewing the QuickML insights

Once the model is integrated, the widget titled Insights from QuickML displays the respective model output in the record detail page.

• To view the insights, click Fetch Insights button.

• Each configuration’s last updated time and refresh option appear at the bottom of the widget.
• If multiple configurations exist for a module, users can switch between them using the arrows in the widget header.
  

Understanding QuickML output in Zoho CRM

QuickML currently supports four types of model outputs, which are displayed as widget variations in CRM:

1. Prediction: Displays the predicted outcome along with the likelihood percentage of the outcome being correct. You can also view the key factors that influenced this prediction by clicking the View Top Influencing Factors.
2. Forecasting: Shows the forecasted value for a selected date.
3. Recommendation: Provides personalized product recommendations for each record and users can view all recommended products by clicking View All Recommendations.
4. Text Analytics: Displays the results of text analysis such as sentiment detection, keyword extraction, etc.

Each output type is visualized within the record’s Insights from QuickML widget.

You can also customize the widget in the Canvas detail view, to align with your business style and requirements.

Pricing

QuickML APIs are billed based on usage. For more details on the pricing, click here.