NexusAI

AI-Powered Sales Intelligence Platform

Multi-agent research pipeline • Company knowledge graphs • Hybrid lead scoring • Opportunity detection

---

System Architecture

graph LR
    A[User Input] --> B[Lead Ingestion API]
    B --> C[Background Job Queue]
    C --> D[Agent Planning]
    D --> E[Parallel Research Workers]
    E --> F{Data Sufficient?}
    F -->|No| E
    F -->|Yes| G[Company Enrichment]
    G --> H[Knowledge Graph Builder]
    H --> I[Hybrid Lead Scoring]
    I --> J[Opportunity Detector]
    J --> K[Outreach Generator]
    K --> L[Frontend Dashboard]

    style A fill:#7c3aed,stroke:#7c3aed,color:#fff
    style C fill:#dc382d,stroke:#dc382d,color:#fff
    style D fill:#0891b2,stroke:#0891b2,color:#fff
    style H fill:#059669,stroke:#059669,color:#fff
    style I fill:#d97706,stroke:#d97706,color:#fff
    style L fill:#7c3aed,stroke:#7c3aed,color:#fff

Loading

Pipeline Workflow

stateDiagram-v2
    [*] --> Plan: Lead submitted
    Plan --> Research: Generate research plan
    Research --> Evaluate: Parallel data gathering
    Evaluate --> Research: Insufficient data (retry)
    Evaluate --> Structure: Data complete
    Structure --> Score: Build company profile
    Score --> Detect: Hybrid scoring
    Detect --> Outreach: Opportunity signals
    Outreach --> [*]: Grounded email draft

Loading

Key Features

Feature	Description
Planning Agent	LLM generates research plan before execution
Parallel Research	6 concurrent research workers via asyncio.gather
Knowledge Graph	Persistent entity-relationship graph per company
Hybrid Scoring	5 deterministic rules (0-20 each) + LLM reasoning
Opportunity Detection	Funding, hiring, tech migration, product launch signals
Grounded Outreach	Every claim maps to a verified research source
Background Jobs	Async queue with real-time SSE progress streaming
Redis Caching	6-hour TTL on research data to reduce API calls
Rate Limiting	slowapi: 50 req/min/IP, 10 research jobs/min
Structured Logging	JSON telemetry with pipeline stage timing

Tech Stack

Backend: FastAPI, LangGraph, Groq (Llama 3.3 70B), SQLAlchemy, Redis, SSE

Frontend: Next.js 15, Tailwind CSS, Shadcn UI, Framer Motion, Recharts, SWR, Clerk

Infrastructure: PostgreSQL/SQLite, Redis (caching + job queue), slowapi rate limiting

Quick Start

Backend

cd backend
python -m venv venv && venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env  # Fill in API keys
python -m uvicorn main:app --reload --port 8000

Frontend

cd frontend
npm install
cp .env.example .env.local  # Fill in keys
npm run dev

Environment Variables

Variable	Required	Description
GROQ_API_KEY	Yes	Groq API key
DATABASE_URL	No	PostgreSQL URL (defaults to SQLite)
REDIS_URL	No	Redis URL (defaults to localhost:6379)
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY	Yes	Clerk auth key

API Endpoints

All routes prefixed with /api/v1/

Method	Route	Description
POST	/leads	Create lead
POST	/leads/{id}/process	Enqueue research job
GET	/companies	List companies with search
GET	/companies/{id}	Full intelligence profile
GET	/companies/{id}/graph	Company knowledge graph
GET	/graph	Global knowledge graph
GET	/jobs/{id}/stream	SSE progress stream
GET	/analytics/pipeline-overview	Pipeline funnel data
GET	/analytics/opportunities	All opportunity signals

Project Structure

nexusai/
├── backend/
│   ├── main.py                    FastAPI app with workers
│   ├── config/settings.py         Environment configuration
│   ├── api/routes/                7 versioned route modules
│   ├── src/
│   │   ├── database/              SQLAlchemy models + operations
│   │   ├── graph/                 LangGraph agent pipeline
│   │   ├── infrastructure/        Redis, queue, telemetry
│   │   ├── services/              6 business logic services
│   │   ├── llm/                   LLM provider abstraction
│   │   ├── models/                Pydantic + SQLAlchemy models
│   │   └── tools/                 Search, scraping, email
│   └── requirements.txt
├── frontend/
│   ├── src/
│   │   ├── app/dashboard/         Pages (overview, leads, companies, graph)
│   │   ├── components/            UI + dashboard components
│   │   ├── hooks/                 SWR data fetching hooks
│   │   └── lib/                   API client, utilities
│   └── package.json
└── README.md

License

MIT