How to Build an AI Content Classifier?

Project Access Link

https://editor.momen.app/tool/rmLyJ0Z0g6o/WEB?code=vOfYHu1Yu7tXe

Introduction

Goal: Automatically classify user feedback using AI and save the results to a database.
Applicable Scenario: Customer service automation, sentiment analysis for support tickets, or automated feedback sorting.
Core Logic: Capture user input through a Text Input -> Trigger a backend Actionflow -> Invoke an AI Agent with a structured output schema -> Insert the analyzed results (category, urgency, sentiment) into the database.

Steps

Data Storage

Create a table to store both the original feedback and the metadata generated by the AI.

Data Model (Table Name: ticket):

Field Name	Type	Note
`description`	Text	Original feedback provided by the user.
`category`	Text	AI-determined category (e.g., Billing Issue, Technical Support).
`urgency`	Text	AI-determined priority level (High, Medium, Low).
`is_positive`	Boolean	Sentiment analysis (True for positive/neutral, False for negative).

Logic & State Configuration

AI Agent Configuration

Navigate to the AI tab and create an AI Agent to handle the classification logic.

Agent Name: Agent_feedback
Inputs: Add an input named description (Type: Text).
Prompt Template: Define clear rules to ensure the AI understands the business logic.
Outputs (Structured):

Tip: You must select Structured output mode. This ensures the AI fills out a predefined “form,” providing valid JSON that the backend can process directly.

Define the following fields:

category (Text, Required)
urgency (Text, Required)
is_positive (Boolean, Required)

Actionflow Construction

Create an Actionflow to connect the UI, AI, and Database.

Actionflow Name: AI Content Classifier
Trigger/Input: Add a text input named description.
Execution Mode: Async (Required for running AI nodes effectively).
Action Steps:
1. AI - Start Conversation:
  - Select Agent: Agent_feedback.
  - Binding: Map description to Actionflow data -> Input -> description.
2. Database - Insert Data:
  - Table: ticket.
  - Field Bindings:
    - description: Actionflow data -> Input -> description.
    - category: AI node -> data -> category.
    - urgency: AI node -> data -> urgency.
    - is_positive: AI node -> data -> is_positive.

UI Construction & Interaction

Component Tree

Page
- Text Input
- Button

Configuration Details

Button Interaction:
- Trigger: OnClick
- Action: Run Actionflow -> AI Content Classifier.
- Data Binding: Bind the description parameter of the Actionflow to Inputs -> Text Input -> Value.

Verification

Preview Checklist:
1. Open Preview mode.
2. Enter “The app keeps crashing” into the Text Input.
3. Click the Button.
4. Navigate to Data Source -> Database -> ticket table.
5. Verify a new record exists:
  - category: “Technical Support”
  - urgency: “High”
  - is_positive: false