24.1 External AI Provider Support
As of February 16, 2026, the Operations module in the Developer Console explicitly supports configuration of the following external AI providers via BYOK approach (February 2026 Release Notes):
- OpenAI (GPT models)
- Anthropic (Claude models)
- Google (Gemini models)
When an external provider is configured, its API key is stored securely and used for the corresponding AI features. The primary use case is unlocking multimodal and advanced reasoning capabilities not yet available in Zoho's own LLM.
24.2 AI Agents in Creator Microservices
AI Agents are a Microservices module component that enables autonomous, multi-step task execution using natural language instructions. An AI Agent in Creator:
- Accepts a natural language task description.
- Associates pre-configured Deluge functions as callable tools.
- Generates an endpoint URL that can be invoked from a Creator application (via the Invoke URL Deluge task or a Widget's JS API).
- At runtime, the agent autonomously analyzes the request, plans an execution sequence, selects the appropriate Deluge functions, and executes them.
This architecture allows Creator applications to expose agentic endpoints that external systems or internal users can call to trigger intelligent, multi-step business processes without explicitly programming the decision logic.
Access: Admin Dashboard → Microservices → + Create New → AI Models → AI Agent.
24.3 Zoho MCP (Model Context Protocol)
Zoho MCP (zoho.com/mcp/) is a separate but closely related Zoho product that standardizes how AI agents interact with Zoho applications (including Creator) and third-party services. It implements the open Model Context Protocol (MCP) standard to expose Zoho app tools, actions, and contextual data in a format any LLM-powered agent can consume (Zoho MCP).
Zoho MCP is not part of Creator directly, but Creator is one of the supported Zoho services exposed via Zoho MCP. This means external AI agents (using Claude, GPT, or any MCP-compatible LLM) can interact with Creator records, trigger functions, and query data through the MCP protocol.
Key Zoho MCP characteristics:
- Model-agnostic: Works with any LLM (GPT, Claude, Gemini, open-source models).
- Low-code configuration: MCP servers can be set up via a UI without extensive coding.
- Schema-first design: Self-describing APIs with input/output contracts.
- Security: OAuth-based authorization with user-level permission scoping.
- 500+ integrations across Asana, Twilio, Zoho, and beyond.
- Autonomous operation: Supports fully autonomous agents that monitor, reason, and act without human prompting.
- Remote MCP Server for Analytics (February 2026): A centrally managed MCP Server over HTTPS for AI agent access to Zoho Analytics data.
The practical effect for Creator is that organizations can build sophisticated agentic workflows where AI agents orchestrate Creator data operations as part of broader multi-app processes—without the agents needing custom API integration code.