Skip to main content

Platform Overview and Development Modes

On the Lovrabet platform, AI can help you rapidly generate complete business systems. But we are also well aware that AI is not omnipotent -- some tasks it handles well, while others require human involvement.

What AI Can and Cannot Do​

Lovrabet's AI can help you build a complete business system (CRM, ERP, order management, etc.) in a matter of days, whereas traditional development might take 2-3 months. The generated code meets enterprise standards, and business users can drive the generation process by describing requirements in natural language.

However, AI is currently not great at these things:

  • Deep industry expertise: The risk models or supply chain optimization algorithms you have built over 10 years are hard for AI to replace
  • Company-specific rules: Your organization's unique approval workflows, organizational structure, and decision-making mechanisms
  • Exceptional user experience: Meticulously crafted interaction design and innovative functional experiences
  • Multi-system integration: Deep integration with existing systems like WeCom, DingTalk, SAP, Yonyou, etc.
  • Breakthrough innovation: Disruptive business models or industry-first features

So our approach is: let AI handle 80% of the standard features, and you focus on the 20% that delivers core value. This is what we call "Integration Development."

When to Use Integration Development​

Here are a few typical examples to help you determine whether your requirements call for integration development.

Industry-Specific Algorithms​

You are a risk control expert at a fintech company with a proprietary credit scoring model containing 200+ risk features. AI generates the standard customer management and loan application features, and you use integration development to connect the scoring engine -- reading customer data via SDK, calling the scoring API, and writing results back to the system.

Multi-System Integration​

Your enterprise has an ERP (Yonyou), a CRM (Salesforce), and WeCom that need to be connected. AI generates a unified data dashboard and operational interface, and you use integration development to integrate with each system's API, consolidating multi-source data on extension pages.

Custom Interactions​

You want to build a drag-and-drop project board (similar to Trello), or a visual supply chain network diagram. AI handles the underlying data storage and permission control, while you introduce UI libraries like react-dnd and G6 to freely design interactions and visual effects.

Complex Business Processes​

The order approval workflow is complex: assign reviewers based on amount, customer tier, and product type; after approval, trigger inventory deduction, financial bookkeeping, and shipping notifications. AI generates the basic order management features, and you implement the approval engine and automation workflows through integration development.

Cutting-Edge Technology Applications​

You want to integrate an AI customer service bot, speech recognition, or image recognition into the system. AI generates the basic business management features, and you connect services from OpenAI, Alibaba Cloud, Tencent Cloud, etc. through integration development.

Main Application and Sub-application Architecture​

Lovrabet adopts a micro-frontend architecture. The AI-generated system serves as the main application, and the features you develop are sub-applications -- both seamlessly integrated.

┌─────────────────────────────────────────────────────────┐
│                    Lovrabet Workspace (Main Application) │
│  ┌─────────────────────────────────────────────────────┐ │
│  │  AI-generated business system (70% standard features)│ │
│  └─────────────────────────────────────────────────────┘ │
│                          ▲                               │
│                          │ Integration                   │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐              │
│  │ Data     │  │ Custom   │  │ Special  │    Sub-      │
│  │ Dashboard│  │ Reports  │  │ Business │    apps      │
│  └──────────┘  └──────────┘  └──────────┘  (30% custom)│
└─────────────────────────────────────────────────────────┘
Main Application (AI-generated)Sub-application (Developer Extensions)
CreationAI-generated automaticallyDeveloped by technical staff
Coverage70% standard features30% custom requirements
Typical FeaturesCRUD, permissions, approvalsData dashboards, complex reports, custom interactions
Learn More

For more technical details about the main application and sub-application architecture, read Main Application and Sub-application Architecture.

Toolchain Overview​

To make integration development smoother, we provide a complete toolchain. Here is a brief overview -- subsequent tutorials will cover each tool in detail.

Why Can AI Handle Development Automatically?

You ask AI to write business code. It generates beautiful frontend components -- but all the data is const mockData = [...], because there is no backend. You have to build the backend first: choose a framework, connect to a database, write CRUD APIs, handle authentication, configure CORS... Building the backend takes a week, and integration testing takes another week (field names don't match, parameter formats don't match, response structures don't match). What about multi-table SQL joins? AI's guessed JOIN conditions are correct 3 out of 10 times -- it doesn't know that the foreign key is customer_id, not customerId. A feature that should take 1 day ends up requiring 2 weeks of frontend-backend integration. The rabetbase trio eliminates this vicious cycle:

  • BaaS (core competency): The database reverse-engineering engine automatically extracts business models, with APIs, SDKs, and enterprise-grade permissions all ready -- no backend setup, no API writing, no integration testing needed. AI works with real business structures, not mock data.
  • CLI: Powerful system-level tools -- covering the full development chain from project create to menu sync. No manual project setup, build configuration, or menu registration needed.
  • Skill: Guides AI to correctly use CLI and BaaS with a standard operating procedure -- field names are correct, API calls are correct, SQL JOIN conditions are correct.

After installing the rabetbase CLI and Skill, you only need to describe your requirements in natural language:

Use rabetbase CLI to help me create a customer management sub-application, including a customer list (with search and pagination), customer detail page (with editing), and a new customer form (with phone number validation).

AI will use rabetbase project create to create the project, rabetbase api pull to generate SDK configuration, then automatically generate all page code, and finally use rabetbase run build + rabetbase menu sync to build and deploy.

Lovrabet SDK -- Data Access​

Without the SDK, you have to study API documentation, write request code manually, and handle authentication yourself. With the SDK, a few lines of code complete data operations:

// Query customer list
const customers = await client.models.customers.filter({
  currentPage: 1,
  pageSize: 20,
});

// Create customer
const newCustomer = await client.models.customers.create({
  name: "张三",
  phone: "13800138000",
  company: "ABC科技",
});

The SDK supports multiple authentication modes, comes with complete TypeScript type definitions, and the same code can run in both Node.js and the browser.

Detailed documentation: Lovrabet SDK Guide

Lovrabet CLI -- Project Scaffolding​

Without the CLI, you have to build projects from scratch, configure menus manually, and handle various integration details. With the CLI:

# Create project with one command
rabetbase project create my-app

# Auto-generate API configuration
rabetbase api pull --appcode your-app-code

# Sync menus to main application with one command
rabetbase menu sync

The CLI automatically scans pages under src/pages, intelligently extracts menu names, and creates menus in batch -- no manual configuration needed, zero omissions.

Detailed documentation: Rabetbase CLI Guide

Rabetbase Skill -- AI-Assisted Development​

Rabetbase Skill enables AI tools (Claude Code, Cursor, etc.) to understand your business data.

Without Skill, when you ask AI to write query code, AI doesn't know your table structure. The field names and API endpoints it generates are likely wrong, requiring repeated back-and-forth to fix.

With Skill, AI automatically retrieves your dataset structure -- table names, field names, and field types are all clear, and the generated code is essentially correct on the first try.

Installation is simple:

npx skills add lovrabet/rabetbase --global

Detailed documentation: Rabetbase CLI Quick Start

Next Steps​

If you have a good understanding of the overall integration development approach, continue reading: