Custom APIs and Microservices

12.1 Microservices Module

The Microservices module is a dedicated section of the Creator admin dashboard that serves as a hub for backend service components. It includes:

• Custom APIs: User-defined API endpoints backed by Deluge functions
• AI Models: Custom-trained and ready-to-use AI models (via the AI Modeler)
• AI Agents: Agentic AI configurations (see Section 23)
• Functions: Reusable Deluge functions

12.2 Custom APIs

Custom APIs allow Creator developers to expose specific Deluge function logic as callable HTTP endpoints from external systems. This creates a powerful reverse-integration capability: external applications and systems can invoke Creator business logic through a standardized API interface without any custom server infrastructure (Understand Custom APIs).

Endpoint format: https://www.zohoapis.com/creator/custom/<appadmin_name>/<customAPIname>

Configuration options:

• HTTP Method: GET, POST, PUT, or DELETE
• Content Type: application/json or multipart/form-data
• Argument Type: Key/Value pairs or Entire JSON
• Response Type: Standard (status codes) or Custom (defined response schema)
• User Scope: Admin Only, Selective Users, Portal Users, or All Users (Portal Users and All Users were added in March 2025)
• Authentication: OAuth (Creator token) or Public Key (API key appended as query parameter)

As of March 2025, Custom API enhancements include:

• Testing custom APIs in environments before deployment
• Portal users can invoke custom APIs via widgets
• Custom APIs usable in environment-enabled functions (March 2025 Upcoming Updates)

12.3 Bulk Insert APIs (April 2026)

Released April 15, 2026, Bulk Insert APIs enable asynchronous, high-volume record insertion into Creator forms. Useful for data migration from external systems. The API set includes (April 2026 Release Notes):

• Create Bulk Insert Job
• Upload File to Bulk Insert Job
• Start / Abort Bulk Insert Job
• Get Bulk Insert Job Status
• Download Bulk Insert Result

The process is asynchronous: a job is created, a file uploaded, the job started, and results polled separately. This enables large-scale data imports without blocking or timeout issues.