Steven3-art/telecom-ai-agent

AI-powered FTTH subscriber support automation for African telecom operators | Qwen Cloud Hackathon (Track 4) + Slack Agent Builder Challenge # TelecomAI Agent 🌍📡 > AI-powered FTTH subscriber support automation for African telecom operators. Built for the **Qwen Cloud Hackathon** (Track 4: Autopilot Agent) and the **Slack Agent Builder Challenge** (Slack Agent for Organizations). ## 🎯 Problem African telecom operators receive hundreds of FTTH support requests daily via email. Each request requires manual portal access, diagnosis, and response — taking 5–15 minutes per case. ## 💡 Solution TelecomAI Agent automates the full support pipeline: ``` Email / Slack request ↓ AI Agent identifies subscriber number ↓ Automated AAA portal consultation ↓ Intelligent diagnosis (6 categories) ↓ Instant response + ETL recording ``` ## 🏗️ Architecture ``` ┌─────────────────────────────────────────────┐ │ TELECOMAI AGENT │ │ │ │ core/ │ │ ├── agent.py ← Main pipeline │ │ ├── aaa_mock.py ← AAA simulation │ │ └── diagnostics.py ← AI categorization │ │ │ │ qwen_interface/ ← Alibaba Cloud │ │ slack_interface/ ← Slack Bot │ └─────────────────────────────────────────────┘ ``` ## 🚀 Quick Start ```bash pip install -r requirements.txt python core/agent.py ``` ## 📊 Diagnostic Categories | Category | Description | Action | |----------|-------------|--------| | NORMAL_EN_LIGNE | Active with traffic | No action needed | | ABSENCE_AUTH | Active in AAA, no connection | CERAF investigation | | ECHEC_AUTH | Authentication failures | Password reset | | SUSPENDU | Account suspended | Commercial contact | | INEXISTANT | Not in AAA | Check contract | ## 🌍 Impact Targeting 45+ million FTTH subscribers across Sub-Saharan Africa. Reduces support response time from 15 minutes to under 30 seconds. ## 📄 License MIT License — Open Source