AI Assistant (Beta)

6 minutes to read

Caspio AI Assistant (Beta) is a platform AI feature that lets you build Flex applications using natural language. You describe what you want to build, and the assistant generates a structured application plan for your review. Once approved, it creates the application automatically. You can continue refining the application through conversation or update it manually in the Flex Designer. Both approaches work together within the same platform.

How it works

AI Assistant (Beta) enables you to create and evolve a Flex application using natural language. Simply describe the application you want to build, and the assistant generates a proposed configuration for your review. You can continue refining the proposal through conversation, making adjustments and exploring alternatives until it meets your requirements. Once you’re satisfied, approve the proposal and the assistant will automatically provision the application’s data model, sample data, directories, roles, segments, AppPages, DataParts, navigation, and inter-page logic in a single step.

After the initial generation, you can continue the conversation to enhance or modify the application. You can also make changes directly in the Flex Designer at any time. AI-assisted development and manual configuration complement each other, allowing you to combine conversational app building with the full flexibility of the designer.

Enabling AI Assistant (Beta)

If you are an account owner or administrator, you can control whether AI Assistant (Beta) is available to users in your account by enabling or disabling the feature in Account Settings.

When enabled, AI Assistant (Beta) is accessible from the header panel throughout the following areas:

Flex application lists
All Flex AppPages in Designer
Tables list

Generate a Flex application

Open AI Assistant (Beta) from the launcher on Flex applications list page.
Describe the application you want to build in natural language. Be as specific as needed. The assistant can generate the data model, pages, navigation, and application logic based on your requirements.
Review the proposed application configuration. The assistant presents a detailed proposal showing what will be created before any changes are applied.
Refine the proposal through conversation as needed. You can continue iterating until the application design meets your requirements.
Approve the proposal to create the application. The assistant provisions all required components in a single operation.
Continue the conversation to extend or modify the application, or use the Flex Designer to make manual changes.

Supported operations

The AI Assistant can perform the following operations:

Applications: Create, rename, and list Flex applications.
Tables: Create and list tables, add fields, and add sample data.
Views: List fields in views.
Directories: Create and list directories.
Segments and Roles: Create and list segments and roles.
AppPages: Create, rename, delete, and list AppPages.
DataParts: Create and list DataParts, add and remove fields, and edit settings. Supported DataPart types include forms, tabular reports, and all chart types except combination charts.
Relationships: Create, rename, delete, and list relationships.
Filtering: Filter DataParts using other DataPart fields.
Interactive Controls: Create, edit, delete, and list controls in DataParts.

Sample prompts

To help you get started and make the most of the power that AI Assistant (Beta) provides, analyze the following sample prompts. You can use these examples as the baseline for your own interactions with AI Assistant:

Sample app creation prompts

Simple app: “Build a project tracker app“
App with context: “Build a project tracker app for a small team that assigns tasks, tracks deadlines, and logs status updates”
Domain specific app: “Build a patient intake app for a medical clinic with appointment scheduling and intake forms“

Sample app object modification prompts

Add a DataPart: “Add a chart that shows total tasks by status“
Add a field: “Add a priority field to the Tasks table“
Rename an AppPage: “Rename the Home AppPage to Dashboard“
Add a role: “Create a Manager role that can see all records“
Add sample data: “Add sample data to the Tasks table so I can preview the app“

You can adapt and modify these prompts based on your specific needs, and then let AI Assistant (Beta) transform your ideas into applications.

Prompt quality

The quality of your prompt is crucial in obtaining the desired response from such AI-powered solutions as AI Assistant (Beta). A clear, specific, and well-informed prompt provides the necessary guidance to steer the AI toward generating a relevant and meaningful output that aligns with your expectations. In general, the better the quality of your prompt, the better aligned the response.

Response probability

Because AI Assistant (Beta) uses an NLP model, it relies on the concept of probability that is central to response generation. The model calculates the likelihood of words and phrases based on the input context, using this probability to determine the most likely and coherent response. It generates responses that are statistically probable based on the given input and the vast amount of training data it has been exposed to.

Response time

Response time depends on the complexity of your prompt and may take up to a few minutes. If no response is returned within that time or an error is shown, it typically means the request is too complex or not supported by Caspio. In such cases, you will be prompted to revise your prompt.

Frequently asked questions (FAQ)

Is my table data accessible by AI Assistant (Beta)?
No, data in your Caspio tables is not accessed. At this time, AI Assistant (Beta) is limited to new object creation based on instructional prompts. To protect sensitive information, do not include it in your prompts.
What AI model does AI Assistant (Beta) use?
AI Assistant utilizes OpenAI as the underlying technology for natural language processing (NLP). OpenAI uses your prompts to generate suggestions for database schema, table relations, and so on.
Is my data used to train AI models?
No, per our signed agreement, data transferred through OpenAI’s API services is not used for model training or to develop AI products. Data is automatically deleted after 30 days.

Resource Usage

Each request consumes Caspio Credits. The number of Credits consumed depends on the complexity of the request. For example, generating a full application with multiple pages consumes more Credits than a simple operation such as adding a field or renaming an object. Credits are consumed when you submit a prompt, regardless of whether you approve the generated output. See Caspio Credits.