---

🎯 Problem Statement

Modern SaaS companies receive hundreds or even thousands of customer support requests, reviews, and bug reports every day. As customer volume grows, manually reviewing every conversation becomes slow, inconsistent, and expensive. This makes it difficult for support and product teams to identify recurring issues, understand customer sentiment, prioritize engineering work, and respond quickly to emerging problems.

Businesses need an intelligent system that can automatically analyze customer conversations, detect recurring complaint patterns, monitor sentiment trends over time, escalate critical issues when necessary, and generate actionable business recommendations. Without these capabilities, valuable customer insights are often missed, leading to slower decision-making, reduced customer satisfaction, and increased operational costs.

SmartSupport AI solves this challenge by using a multi-agent architecture to automate customer support, continuously analyze customer feedback in real time, and transform customer conversations into actionable business intelligence through an interactive analytics dashboard.

---

💡 Solution

SmartSupport AI is a production-ready B2B Multi-Agent Customer Support & Customer Intelligence Platform built with Google ADK, MCP, and Gemini.

The platform transforms customer conversations into actionable business insights by combining intelligent support automation with real-time analytics.

It enables businesses to:

• 🤖 Automatically classify customer issues and determine whether they can be resolved instantly or require technical escalation.
• 💬 Generate professional, empathetic support responses to improve customer experience.
• 📊 Detect recurring complaint patterns before they become major business problems.
• 😊 Analyze customer sentiment and monitor trends over time.
• 📈 Generate prioritized business recommendations that help product and support teams focus on the highest-impact issues.
• 📉 Visualize insights through an interactive dashboard with complaint trends, sentiment analytics, and business KPIs.

Instead of manually reviewing hundreds of customer conversations, SmartSupport AI enables six specialized AI agents to analyze issues in seconds and provide clear, actionable insights for faster business decisions.

---

🏗️ Architecture

---

🤖 Agents (Google ADK + Gemini 2.5 Flash)

Agent	Role
Support Agent	Classifies complaints, drafts replies, decides escalation
Orchestrator Agent	Coordinates all analysis agents in pipeline
Complaint Identifier	Finds top repeated complaints, groups by category
Sentiment Analyzer	Scores each issue, tracks monthly sentiment trends
Trend Analyzer	Detects growing issues, flags worsening problems
Insight Reporter	Generates prioritized business recommendations

---

🔧 MCP Tools (FastMCP Server)

Tool	Purpose
save_issue()	Save customer issue to SQLite with session ID
fetch_all_issues()	Retrieve all issues from database
fetch_active_issues()	Get only Open/In Progress issues for analysis
update_issue_status()	Mark issue as Resolved/In Progress + timestamp
get_repeat_issues()	Detect recurring complaints by category
save_dashboard_data()	Save agent analysis results for dashboard
create_session()	Generate unique session ID for conversation tracking
get_conversation_history()	Retrieve full conversation history by session
save_conversation_message()	Save each customer/agent message to database

---

✨ Key Features

🧠 Smart Escalation Logic

self_fixable → Agent gives troubleshooting steps Issue saved as "Pending Customer Action" Orchestrator NOT triggered yet technical_escalation → Agent escalates immediately Issue saved as "Open" Orchestrator triggered instantly Business dashboard updated

📊 Business Intelligence Dashboard

• Real-time sentiment trend charts
• Top complaint categories bar chart
• Daily issue volume tracking
• AI-generated business recommendations
• Issue management with status updates
• Recurring issue alerts

💬 Customer Support Portal

• WhatsApp-style chat interface
• Instant AI replies
• Smart classification badges
• Escalation status indicators

---

🛠️ Tech Stack

Category	Technology
AI Framework	Google ADK 1.3.0
LLM	Gemini 2.5 Flash
Backend	FastAPI + Uvicorn
MCP Server	FastMCP
Database	SQLite3
Frontend	HTML5 + CSS3 + JavaScript
Charts	Chart.js
Package Manager	pip
Language	Python 3.13

---

📁 Project Structure

customer_insight_agent/
├── .env                          # API keys (not in git)
├── requirements.txt              # Dependencies
├── api/
│   └── server.py                 # FastAPI backend
├── agents/
│   ├── complaint_identifier.py   # Finds repeated complaints
│   ├── sentiment_analyzer.py     # Analyzes sentiment trends
│   ├── trend_analyzer.py         # Monthly trend analysis
│   ├── insight_reporter.py       # Business recommendations
│   └── support_agent.py          # Smart reply + escalation
├── orchestrator/
│   └── orchestrator_agent.py     # Pipeline coordinator
├── mcp_tools/
│   └── data_tool.py              # MCP Server (6 tools)
├── frontend/
│   ├── dashboard.html            # Business dashboard
│   └── customer.html             # Customer chat portal
├── data/
│   └── issues.db                 # SQLite database
└── README.md

---

⚙️ Setup

Prerequisites

• Python 3.13+
• Gemini API Key from Google AI Studio

Installation

# 1. Clone the repository
git clone https://github.com/abubakar1yousafzai/SmartSupport-AI
cd SmartSupport-AI

# 2. Install dependencies
pip install -r requirements.txt

# 3. Create .env file
echo "GEMINI_API_KEY=your_key_here" > .env

# 4. Start the server
uvicorn api.server:app --host 0.0.0.0 --port 8000 --reload

# 5. Open in browser
# Business Dashboard: http://localhost:8000
# Customer Portal:    http://localhost:8000/customer
# API Docs:           http://localhost:8000/docs

---

📹 Video Demo

---

🎓 Course Concepts Demonstrated

Concept	Implementation
Multi-Agent System (ADK)	6 specialized agents with orchestrator
MCP Server	FastMCP with 9 tools
Security	API keys in .env, CORS, input validation
Deployability	FastAPI + Uvicorn, one command setup
Antigravity IDE	Used for development and agent building

---

📄 License

MIT License — feel free to use and modify.

---