1. Overview

1. The configured Large Language Model (LLM) processes the input.
   
2. Zia applies application metadata such as forms, fields, and relationships to contextualize the response.
   
3. The refined output is presented within Creator for review and implementation.
   

1. Building applications through conversational prompts
2. Creating forms using natural language descriptions
   
3. Suggesting relevant fields and providing next field recommendations
   
4. Generating Deluge scripts aligned with application context
   
5. Enabling AI Agents to automate tasks
   
6. Performing intelligent data operations using the Deluge Zia task
   

1.1. Building applications through conversational prompts

1. Provide a text prompt describing your application requirements, such as the business problem to be solved or the workflow you want to support.
   
2. Optionally upload relevant documents (e.g., PRDs, process diagrams) to give additional context to the prompt and improve the AI’s understanding.
   
3. Review and refine the generated use cases by adding custom use cases, remove irrelevant ones, or generate additional suggestions.
   Learn how to write effective prompts for Zia App Builder

1.2. Creating forms using natural language descriptions

1.3. Suggesting relevant fields and next field suggestions 

1. Understands the context of the form and existing fields.
   
2. Suggests appropriate additional fields with suitable types.
   
3. Helps maintain logical data structure.
   

1.4. Generating Deluge scripts with prompts 

1. Zia interprets the described requirement.
   
2. Analyzes application metadata such as forms, fields, and relationships.
   
3. Generates Deluge scripts aligned with the application structure.
   
4. Provides editable script output within the Deluge editor
   

1.5. Enabling AI Agents to automate tasks

Security tip

AI agents are designed to automate and streamline tasks, but they do not operate with unrestricted autonomy. Their behavior is guided entirely by the instructions and tools configured for the AI during setup and they act within the roles and permissions of the user.

​

Make sure AI agents are granted only the permissions required for their intended tasks. This ensures they perform only the actions they are explicitly enabled to execute and do not independently access data or systems beyond their defined scope. 

1.6. Intelligent data operations using the Deluge Zia task

2. Business use cases

Case 1: Accelerated application creation

Let's assume that you're a business analyst who wants to develop a Loan Approval application rapidly, but might not be well-versed with sufficient coding skills. You can create an AI-assisted application with Zia.

To do this, you can access your Zoho Creator account and specify a prompt as follows: "Create an application for managing loans and approvals". As an appropriate response,  Zia will generate relevant use cases and also help translate your vision into a fully-functional application with the necessary forms and field relationships to kickstart your application development.

‌Case 2: Simplified form creation 
Let's assume that you want to create a form for the above Loan Approval application. In general, contextual form creation for specific cases is time-consuming and error-prone. To create a form quickly, you can access your application and specify a prompt as follows: "Create a form to manage loan agents". As a response, Zia will generate 10 relevant form fields that are contextually relevant to the described form. You can select the required form fields and proceed to form creation. 

‌Case 3: Intelligent field suggestions
 In the above use case, Zia has assisted in generating the initial set of fields for the Manage Loan Agents form. You might struggle in adding new relevant fields. This is where Zia provides real-time suggestions for the next relevant form fields by understanding the context and thereby enhancing form building experience. Learn how

Case 4: Automate tasks with AI Agents

A customer has submitted a loan application, which needs to go through processes like verifying applicant details, checking credit score, and assigning a reviewer. Manually handling these steps is tedious and time-consuming. To automate these tasks, you can create an AI agent with a simple instruction like:

"You are a loan verifying agent. You perform the below actions:

    - Verify applicant details and check repaying capacity

    - Check if the credit score is more than 700

    - Assign a reviewer based on the type of loan"

 

Once an agent is created with the above instructions, it acts like a loan agent.  It can be called whenever an application is submitted, and it executes all the mentioned tasks.

 

Case 5: Instant Deluge script to calculate repaying capacity

To verify the loan application, you need to calculate the applicant's repaying capacity. This must be written in a Deluge script. Making it easier, you can write a prompt in Deluge Zia Assistance "Write a script to calculate repaying capacity based on the salary and the location". It will generate the required Deluge script, which can be used in your application.

Case 6: Extract details from the document uploaded using Deluge Zia task

As a loan applicant submits their documents, going through each page and getting the accurate details needs significant attention. To make this effortless, you can use the Zia task in Deluge, which will extract the details and provide you with a structured response.

3. Navigation guide

3.1 Guide to create Zia assisted applications

To create apps using Zia App Builder, ensure that you've configured Zia with GenAI in Zia settings and toggle the Create Application button.

Navigate to the Solutions page and click the + Create Solution button at the top-right corner. The screen will show options to create applications (solutions). Hover on the Applications card and click Select. Next, click Create in the Create using Zia card and provide a brief description of the application you want to create.

3.2 Guide to create Zia assisted forms

To create forms using Zia, ensure to configure your required API key in Zia Settings and toggle the Create Form button.

Navigate to the Solutions page and choose the required application in which you want to create forms using Zia's assistance. Create a form by selecting the Using Zia card beneath the On my own section. Provide a brief description of the form you want to create.

3.3 Guide to generate Zia assisted field suggestions

To generate field suggestions using Zia, ensure to configure your required API key in Zia Settings and toggle the Next Field Suggestions button.

Open the form in which you want to generate field suggestions using Zia's assistance. In the respective form builder, click the Zia icon to generate relevant field suggestions.

3.4 Guide to create AI Agents  

Navigate to Microservices and click the +Create New button in the top-right corner. Choose Create in the AI Models card and select the AI Agent card under the Build - Custom Models section. The AI Agent builder will appear. You can create a new AI Agent as explained in this page.

 

3.5 Guide to generate Deluge scripts using Zia

Go to the required app and create a workflow. Click Add New Action to open the Deluge editor. Click the Zia icon  in the top-right corner, which opens the Zia Assistant window. Here, you can enter a prompt to generate the script.

4. Points to note

1. Zia does not consume any AI calls or API calls from Zoho Creator. However, API rate limits may still apply. These limits are restrictions that the LLM provider imposes on the number of times a user or client can access their services within a specified period of time and vary based on their usage tier and the model consumed.
   
2. Rate limits for LLM providers can be referred by their respective official documentation.
   
1. OpenAI rate limits
   
2. Google Gemini rate limits
   
3. Anthropic rate limits
   
3. When multiple AI features are used in quick succession, the configured provider’s rate limit may be exceeded. This can result in errors or temporary interruption of AI powered functionality.
   

1. Currently, prompts for Zia are supported only in English when Zoho GenAI is selected as the LLM provider. To use other languages, configure Zia with external LLM vendors.
   
2. Sample data generated for each use case while generating the application with Zia illustrates real-world scenarios but uses fictitious information.
   
3. AI-generated content may contain inaccuracies. Verify that all description prompts are accurate and appropriate before use.
   
4. The Zia AI Agent feature is supported only with OpenAI LLM provider.
   
5. AI features can be enabled or disabled only by the super admin, except for the Create Applications feature, which can be managed by both the super admin and admin.
   

1. For Zia features, Zoho Creator uses system prompts or user-defined instructions that combine the user prompt with application metadata to generate contextual responses or automate tasks. All supported GenAI providers, including Zoho GenAI, OpenAI, Google, and Anthropic, are committed to maintaining data privacy.
   
2. It is recommended not to include sensitive or confidential information in prompts while using AI features.
   
3. Zoho GenAI is Zoho’s in house LLM and is available at no additional cost. When using Zoho GenAI, your data remains protected within Zoho’s ecosystem.
   

1. AI-generated content may occasionally contain inaccuracies. It is important to review and validate generated outputs before implementing them in your application.
   
2. When entering prompts, it is recommended to use clear, everyday language to describe what your app or form should collect, track, list, or manage. For example:
   
1. “Create an app to track inventory”
   
2. “Track sales leads”
   
3. “Create a form to collect employee details”
   Learn how to write effective prompts for Zia App Builder
3. Using simple and descriptive prompts improves the accuracy and contextual relevance of generated outputs.
   
4. The sample data provided in use cases represents realistic business scenarios but uses fictitious information for illustration purposes. 
   

5. Glossary

1. Large Language Model (LLM) - An AI model trained on extensive textual data to understand context, semantics, and intent, and to generate coherent, contextually relevant responses. It serves as the underlying technology for natural language understanding and generation within AI-powered features within Creator.
   
2. Zia - Zia is Zoho Creator’s AI assistant layer that powers a wide range of AI capabilities across the platform, including generative AI, machine learning models, and intelligent automation. It integrates underlying technologies such as LLMs and applied AI models to enable features like prompt-based app creation, data processing, predictive analysis, and AI-assisted development workflows.
   
3. Zoho GenAI - Zoho GenAI is Zoho’s native Large Language Model (LLM) that powers AI capabilities within Zoho Creator. It is integrated by default in Zoho Creator and does not require an external API key. It allows users to access AI features while keeping data within Zoho’s ecosystem.
   
4. Bring Your Own Key (BYOK)-  BYOK is a configuration approach that allows users to connect external LLM providers by supplying their own API keys. This enables the use of third-party AI services such as OpenAI, Google, or Anthropic within Zia.
   
5. Metadata - Metadata refers to the structural information of the application, such as forms, fields, relationships, and configurations. Zia uses this context to generate more accurate and relevant outputs.
   

6. Related topics

1. Understand Zia features in Creator
   
2. Creating applications using Zia App Builder
   
3. Create forms using Zia assistance
   
4. Generate field suggestions for AI assisted forms
   
5. Understanding AI Agent
   
6. Deluge Zia Task
   
7. Configuring next field suggestions