How to Build APIs, FaaS Functions, and Direct DB Connections in Baidu’s Low‑Code Platform

1. Review

In the previous part we described the classic single‑API data‑connection method in low‑code platforms and related techniques such as data mapping, adapters, and API proxies that aim to reduce the cost for business users to integrate with the platform.

We also noted that the single‑API approach still depends on existing backend APIs; when those APIs do not meet the needs of low‑code applications, traditional backend development is still required.

This article focuses on scenarios where the backend has no ready‑made API and introduces Baidu Intelligent Cloud’s low‑code platform “AiSuDa” solutions for API orchestration, FaaS, and direct database connection.

2. API Orchestration

API orchestration is a visual way to compose backend logic, useful for simple interface aggregation and CRUD scenarios when a single ready‑made API is unavailable. It lets platform users combine existing APIs into a new composite API without involving traditional backend developers.

The creation and usage process in AiSuDa is as follows:

Create a “Multi‑API Aggregation” entry in the API Center (i.e., an API orchestration).

Enter the visual editor via “Multi‑API Aggregation Design” to arrange API nodes.

Configure components on the page to use the created API orchestration.

2.1 Creating an API Orchestration

In the API Center you can create two types of APIs: single APIs (covered in the previous article) and multi‑API aggregation APIs, which are API orchestrations.

2.2 Visual Editor for API Nodes

The visual editor shows a basic “Send HTTP Request” node, which can be configured to call external services. Additional atomic capabilities such as SQL data source nodes and JavaScript code execution nodes are also available, allowing users to compose new APIs for front‑end consumption.

A typical orchestration retrieves a temporary access token using an AK/SK pair, then uses that token to call a face‑registration API.

The key nodes are:

Exchange AK/SK for a temporary token: three query parameters (grant_type, client_id, client_secret) are sent; the response is stored in a variable accessToken.

Attach the obtained token to the next request as a query parameter and submit the image payload from the front‑end ( {{ input.image }}).

The final output is stored in output.data, so the API response contains a data field with the face‑registration result.

2.3 Using the API Orchestration in a Page

After the orchestration is created, it is exposed as a single endpoint (e.g., /apicenterproxy/:apiId). Users can select the API orchestration in the page editor and bind it to a form submission.

When the form is submitted, the backend engine executes the orchestration, retrieves the token, calls the face‑registration service, and returns the result in the data field.

3. FaaS

Function as a Service (FaaS) lets users upload source code (or paste it) and expose it via triggers, most commonly HTTP. In AiSuDa, FaaS is integrated as a node type within API orchestration.

4. Direct Database Connection

The direct‑connection solution covers three aspects: data‑source access, modeling capability, and data access in low‑code pages.

4.1 Data‑Source Access

Users provide database credentials to the platform, which then connects to the database for CRUD or DDL operations. AiSuDa supports external data sources and built‑in data sources.

4.1.1 Connecting External Data Sources

After submitting the connection details, the platform maps original database tables and fields to platform data models.

4.1.2 Creating Built‑In Data Sources

Select “Built‑in Data Source” when adding a new data source to create an application‑specific database.

The created database appears in the platform’s database list.

4.2 Modeling Capability

4.2.1 Visual Model Editor

The visual editor allows users to create and modify data models (tables) via an ER diagram, including advanced fields that bind to organizational information.

4.2.2 Synchronization Modes

Three modes are supported: No Sync, Automatic Sync, and Manual Sync. In No Sync the model is read‑only; Automatic Sync applies changes to the target database immediately upon saving; Manual Sync generates the SQL diff for DBA review before execution.

Example of Automatic Sync: adding a new field in the visual editor (

), then verifying the field appears in the actual database (

).

5. Data Access in Business Pages

5.1 Model Components

AiSuDa provides two front‑end components: “Model Form” and “Model List”, which bind to a data model and automatically generate UI based on field types.

5.1.1 Model Form

After creating a data model (e.g., fields “Material ID” and “Remarks”), dragging a Model Form component onto a page and binding it generates a form with appropriate input controls.

Submitting the form stores the data, which can be viewed in the Data Management view.

5.1.2 Model List

The Model List component provides full CRUD capabilities for the bound model.

5.2 Accessing Models via API Orchestration Nodes

API orchestration offers SQL nodes and database CRUD nodes to interact with the platform’s managed data models without writing code. The SQL node allows custom queries, while the CRUD nodes provide visual configuration. Additionally, JavaScript nodes (FaaS) can perform database operations.

6. Summary

Dynamic web pages mark the shift from Internet 1.0 to 2.0, and today most front‑end work revolves around handling dynamic data. Connecting to backend data is a major source of friction between front‑end and back‑end teams. Baidu Intelligent Cloud’s low‑code platform AiSuDa offers multiple data‑connection methods—single API, API orchestration, FaaS, and direct database connection—allowing many previously backend‑only requirements (such as data format alignment or custom data retrieval) to be realized through visual configuration, thereby reducing development and communication costs.