OutreachAI – Cold Email Agent

A full-stack AI-powered cold email agent. Upload your resume, enter a recipient's details, optionally trigger real-time research via Perplexity Sonar, and generate + send a personalized cold email — all from a clean React UI.

---

Tech Stack

Layer	Technology
Frontend	React 19, Vite, Tailwind CSS
Backend	Python, FastAPI
Email Generation	Groq API – Llama 3.3-70B Versatile
Recipient Research	Perplexity Sonar API (real-time web search)
Resume Parsing	PyPDF2 (PDF), plain text fallback
Email Delivery	Gmail SMTP via App Password
Database	SQLite (outreach logging)

---

Features

• Resume upload — Drag-and-drop or click to upload PDF, TXT, or MD (max 10MB); text is extracted and injected into the prompt
• Recipient research toggle — Optional checkbox that calls Perplexity Sonar to pull real-time facts about the recipient and company before generating
• AI-generated emails — Groq (Llama 3.3-70B) writes a personalized subject + body grounded in your resume, research context, job link, LinkedIn, and GitHub
• Prompt engineering — Dedicated prompt.py with a structured system prompt that enforces crisp, role-specific outreach
• Live preview — Review the generated email in EmailPreview.jsx before sending
• One-click delivery — Send directly via Gmail SMTP from the UI
• Outreach logging — Every send attempt (recipient, company, role, subject, body, status) is logged to a local SQLite DB via db.py
• Configurable API base URL — Frontend reads VITE_API_BASE_URL from .env for easy staging/prod overrides

---

Project Structure

OutreachAI/
├── back/
│   ├── app.py              # Consolidated FastAPI app (Perplexity + Gmail SMTP routes)
│   ├── main.py             # Original modular FastAPI entry point
│   ├── generator.py        # Groq email generation logic
│   ├── mailer.py           # Gmail SMTP send function
│   ├── db.py               # SQLite logging (insert + query)
│   ├── prompt.py           # SYSTEM_PROMPT + build_user_prompt()
│   ├── resume_parser.py    # PyPDF2-based PDF + text extraction
│   └── emails.db           # Auto-created SQLite database
└── front/
    └── src/
        └── components/
            ├── EmailForm.jsx       # Input form, resume upload, research toggle
            ├── EmailPreview.jsx    # Preview generated email before sending
            └── SentConfirmation.jsx # Post-send success screen

---

Getting Started

Prerequisites

• Python 3.11+
• Node.js 18+
• Groq API key
• Perplexity API key
• Gmail account with 2FA enabled + an App Password

Backend Setup

cd back
python3 -m venv venv
source venv/bin/activate
pip install fastapi uvicorn groq perplexityai pdfplumber PyPDF2 python-dotenv pydantic

Export environment variables in the same terminal you run uvicorn from:

export GROQ_API_KEY="your_groq_api_key"
export PERPLEXITY_API_KEY="your_perplexity_api_key"
export GMAIL_USER="your_gmail@gmail.com"
export GMAIL_APP_PASSWORD="your_16char_app_password"

Start the server:

uvicorn app:app --reload

API runs at http://localhost:8000. Swagger docs at http://localhost:8000/docs.

Frontend Setup

cd front
npm install

Create front/.env:

VITE_API_BASE_URL=http://localhost:8000

Start the dev server:

npm run dev

App runs at http://localhost:5173.

---

API Endpoints

GET /

Health check. Returns { "status": "running" }.

---

POST /parse-resume

Upload a resume file and get back extracted plain text.

Request: multipart/form-data with a file field (PDF, TXT, or MD, max 10MB)

Response:

{ "resume_text": "..." }

---

POST /research

Uses Perplexity Sonar to research the recipient and company in real time.

Request body:

{
  "name": "Jane Smith",
  "email": "jane@company.com",
  "company": "Acme Corp",
  "role": "Head of Engineering",
  "job_url": "https://company.com/careers/job-id"
}

Response:

{ "research": "..." }

---

POST /generate-email

Generates a personalized cold email using Groq, grounded in resume + optional research.

Request body:

{
  "name": "Jane Smith",
  "email": "jane@company.com",
  "company": "Acme Corp",
  "role": "Head of Engineering",
  "target_role": "Software Engineer",
  "resume_text": "...",
  "research_summary": "...",
  "job_link": "https://...",
  "linkedin": "https://linkedin.com/in/...",
  "github": "https://github.com/...",
  "sign_off": "Best regards",
  "context": "..."
}

Response:

{ "subject": "...", "body": "..." }

---

POST /send-email

Sends the email via Gmail SMTP and returns send status.

Request body:

{
  "to_email": "jane@company.com",
  "subject": "...",
  "body": "..."
}

Response:

{ "status": "sent" }

---

Environment Variables

Variable	Description
GROQ_API_KEY	Groq API key for LLM email generation
PERPLEXITY_API_KEY	Perplexity Sonar API key for recipient research
GMAIL_USER	Gmail address to send from
GMAIL_APP_PASSWORD	Gmail App Password (16 chars, no spaces)
VITE_API_BASE_URL	Frontend base URL for API calls (e.g. http://localhost:8000)

---

License

MIT