AI-Powered SDR System

Intelligent lead qualification and outreach automation using Grok AI. Built to streamline B2B sales pipeline management through adaptive scoring and context-aware messaging.

Core Features

• Intelligent Lead Scoring: Multi-factor assessment (company size, job title, industry fit, engagement) with customizable weights
• Context-Aware Messaging: Pipeline stage-aware generation with 6 message types
• 7-Stage Pipeline: Automated progression from New → Qualified → Contacted → Meeting → Negotiation → Closed
• Activity Audit Trail: Complete timeline of lead interactions and scoring decisions

Tech Stack

Backend: FastAPI, SQLAlchemy, SQLite, Pydantic
Frontend: React, TailwindCSS, shadcn/ui
AI: Grok API (xAI) with custom prompt engineering

Architecture

┌──────────┐      ┌──────────┐      ┌─────────┐
│  React   │ ───► │ FastAPI  │ ───► │ Grok AI │
│  Client  │ ◄─── │  Server  │ ◄─── │   API   │
└──────────┘      └──────────┘      └─────────┘
                        │
                        ▼
                  ┌──────────┐
                  │  SQLite  │
                  │    DB    │
                  └──────────┘

Installation

# Backend
cd backend
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
echo "GROK_API_KEY=your_key" > .env
uvicorn app.main:app --reload --port 8001

# Frontend
cd frontend
npm install && npm run dev

Access at: http://localhost:5173

API Endpoints

POST   /api/leads                 # Create lead (auto-scores)
GET    /api/leads                 # List all leads
PUT    /api/leads/{id}            # Update lead
DELETE /api/leads/{id}            # Soft delete
POST   /api/leads/{id}/score      # Re-score lead
POST   /api/leads/score-batch     # Score all leads
POST   /api/leads/{id}/generate-message  # Generate message
GET    /api/analytics/pipeline    # Pipeline statistics

Full API docs: http://localhost:8001/docs

Performance

• Lead scoring: ~800ms per lead (Grok API latency)
• Batch scoring: Parallel processing with progress updates
• Message generation: ~1.2s average
• Database: Indexed queries, supports 10K+ leads

See benchmarks/ for detailed metrics.

Prompt Engineering

Stage-aware context ensures messages adapt to relationship status:

# Negotiation stage - no product introduction
stage_contexts = {
    "negotiation": "Focus on ROI, pricing, addressing concerns. 
                   They already know the product - don't reintroduce."
}

Evaluation

Model performance tracked via:

• Score consistency tests (±5 point variance)
• Message quality assessment (personalization, tone, CTA)
• Stage-awareness validation
• A/B testing framework (planned)

Future Roadmap

• Email integration (SendGrid/SMTP)
• CRM connectors (Salesforce, HubSpot)
• Advanced analytics dashboard
• Multi-channel outreach (LinkedIn, SMS)

---

License: Proprietary | Built with: Grok AI