Multi-Agent Interactive Assistant

TechXDemo.1.mp4

Overview

A sophisticated multi-agent AI system designed to assist with restaurant bookings, scheduling, and research tasks using specialized agents.

Prerequisites

• Node.js (v16+)
• npm
• API Keys

Installation

1. Clone Repository

git clone https://github.com/TuneHQ/multi-agent-demo
cd multi-agent-demo
npm install

2. Environment Configuration

Create a .env file with these variables:

MODEL_API_KEY=your_api_key_here
MODEL=model_name_here
SERPER_API_KEY=your_serper_api_key
ZOMATO_COOKIE=your_zomato_cookie

API Configuration

• Preferred: TuneStudio

  • API Key: Obtain from TuneStudio
  • Model: openai/gpt-4o-mini
  • Keep baseURL: "https://proxy.tune.app"

• Alternative: OpenAI

  • API Key: Get from OpenAI
  • Model: gpt-4o-mini
  • Remove baseURL in OpenAI client configuration

3. Obtain Required API Keys

• MODEL_API_KEY:
  • TuneStudio
  • OpenAI
• SERPER_API_KEY: serper.dev
• ZOMATO_COOKIE: From Zomato (for restaurant-related functions)

Running the Assistant

npm start

Agents Overview

1. Concierge Agent

• Restaurant recommendations
• Table bookings
• Nearby restaurant search
• Slot availability checking

2. Scheduler Agent

• Meeting scheduling
• Calendar management
• Appointment reminders
• Time slot suggestions

3. Researcher Agent

• In-depth topic research
• Information analysis
• Report generation
• Source verification

4. Manager Agent

• Multi-task coordination
• Context maintenance
• Inter-agent communication
• Complex task decomposition

Interaction Guide

• Use natural language commands
• Specify agent or task requirements
• Type "exit" to end session

Example Interactions

You: Find a restaurant near me
Concierge: Here are nearby restaurants...

You: Schedule a meeting with my team
Scheduler: When would you like to meet?

You: Research AI trends in 2024
Researcher: Analyzing current AI landscape...