Showcase: My Multi-Agent AI System Built in callin.io Using Sub-Workflows

Hello,

I've put together this to help illustrate:

This is my primary agent, which has access to sub-workflows containing AI agents. It functions as the central control, determining which sub-agent should handle a given request.

Next, I link the other agents via tools to the main agent. Each call is directed to these:

• ResearchAgent
  This agent initiates another AI agent workflow with specific research instructions (e.g., “look up current social media trends for skincare” or “explain how RAG works”).
  The main agent forwards the original query along with a clear objective for the sub-agent.

• Agent Confirm
  This agent has access to administrative tools and can execute workflows, such as the GitHub backup. The instruction is always explicit, for example: "Run the GitHub backup workflow in agent confirm". The agent strictly follows instructions and does not make assumptions.

• DocumentAgent
  This agent is designed for file operations. It locates files on Google Drive, retrieves them, and then stores them in Supabase for RAG-style access. It's paired with another AI agent that receives instructions like: "Analyze this invoice and return the total due and due date." or "Answer: What company sent this invoice, and what is the payment method listed?"

All sub-agents operate strictly based on the clear input provided to them, without making any assumptions.

The main agent consistently provides specific context and objectives for each task.

Each agent resides in its own sub-workflow, equipped with its own callin.io agent and tool configuration.

These trigger various actions, such as:

• ResearchAgent returning structured data.
• Agent Confirm executing workflows.
• DocumentAgent retrieving documents and answering questions about them.

The main agent's role is purely delegation: it routes tasks to the appropriate tool and returns the output once completed.
It never acts independently, only directing tasks and relaying results.

This setup is not exhaustive but demonstrates a viable structure for organizing agents.

I hope this is helpful.