Skip to content

README.md

Latest commit

 

History

History
109 lines (72 loc) · 3.46 KB

README.md

File metadata and controls

109 lines (72 loc) · 3.46 KB
Raw

🤖 Personal AI Multi Agent Assistant

A modular multi-agent assistant built with n8n that processes text and voice requests, intelligently routes tasks to specialized agents, and executes actions across email, calendar, research, and support domains. Designed for scalability, contextual understanding, and practical productivity.

📸 Workflow Overview

Workflow screenshots:

Workflow Diagram

Workflow Diagram

Workflow Diagram

Workflow Diagram

Workflow Diagram

🧠 Core Architecture

The system consists of two key layers:

1. Main Agent (Orchestrator)

  • Processes incoming Telegram messages (text or audio)
  • Converts voice to text
  • Analyzes user intent using AI models (e.g., Gemini or OpenAI)
  • Identifies single or composite requests and routes them to appropriate child agents

2. Child Agents (Modular Workflows)

Each child agent is an independent workflow for specific tasks. The main agent activates them as needed.

🧩 Child Agent Descriptions

📧 Email Agent

  • Validates recipient contacts using a Google Sheets database
  • Composes and sends emails with AI-generated content
  • Supports dynamic subjects, message bodies, and optional reminders
  • Can be extended for attachments, CC/BCC, and templates

Example Query:

"Send John an email confirming meeting details."

📅 Calendar Agent

  • Schedules events and meetings with timezone awareness
  • Validates participants through contact database unless explicitly provided
  • Can be extended for recurring events, reminders, and synchronization

Example Query:

"Schedule a call with Sarah next Tuesday at 3 PM."

🛠️ Customer Support Agent (RAG)

  • Uses Retrieval-Augmented Generation to answer questions
  • Pulls information from knowledge base (FAQs, documentation)
  • Ensures accurate responses without hallucinations
  • Responses are delivered unaltered for reliability

Example Query:

"What is the refund policy for premium users?"

🔍 Research Agent

  • Performs online searches based on user queries
  • Provides concise relevant insights with sources
  • Can be extended for deeper analysis and reports

Example Query:

"Find top 3 AI workflow tools in 2025."

🛠️ Technology Stack

  • n8n – Workflow orchestration
  • Telegram Bot – Request channel (text + voice)
  • OpenAI / Gemini – Intent analysis and agent logic
  • Google Sheets – Contact database
  • Gmail / Calendar API – Action execution
  • RAG Stack – Knowledge retrieval for support
  • Web Search API – Research backend

🎯 Use Cases

  • Personal productivity assistant
  • AI-powered customer support bot
  • Internal team assistant
  • Scalable platform for enterprise automation

📌 Notes

  • The system is modular - new agents can be added without core changes
  • Voice requests are processed like text
  • Contact validation ensures task execution accuracy
  • RAG agent responses are based on source materials without alterations