AI GTM Agent

AI GTM Agent is an autonomous Go-To-Market system designed to automate and optimize the outbound sales lifecycle. It leverages agentic workflows to handle lead ingestion, deep research, personalized email drafting, and intelligent reply classification—all while keeping a human in the loop for final approval.

👉 View Integration Roadmap

✨ Key Highlights

• 🤖 Autonomous Research: The ICPPersonaAgent analyzes prospect companies and products to determine fit and key talking points.
• ✉️ Hyper-Personalization: The EmailGeneratorAgent crafts unique, relevant emails based on research data.
• � RAG Knowledge Base: Integrated Retrieval-Augmented Generation (RAG) using ChromaDB to retrieve relevant case studies, pricing, and product info during email generation.
• �🛡 Enterprise-Grade Risk Control: Built-in daily Send Caps, Throttling (delay between sends), and Idempotency checks to prevent spam flags.
• 👮 Human-in-the-Loop: A dedicated Admin UI (served via FastAPI) allows Ops teams to batch-review, edit, and approve drafts.
• 🔬 Full Observability: Complete event lifecycle logging (INGEST -> GEN -> APPROVE -> SEND) for audit trails.
• 📥 Smart Inbox Management: ReplyClassifierAgent automatically categorizes replies and stops follow-ups for interested/unsubscribed leads.

🚀 Quick Start (The "Magic Moment")

We have included a pilot onboarding script to demonstrate the full agent capability in 30 seconds.

1. Start the Backend Server:

   python3 -m backend.api.server

2. Run the Pilot Demo (in a new terminal):

   python3 scripts/pilot_onboarding.py

   Follow the interactive prompts to define your ICP, generate test leads, and launch your first AI campaign.

3. Visit the Dashboard: Open http://localhost:8000 to verify the metrics and logs.

🏗 Architecture

The system follows a modular, event-driven architecture organized into specialized agents and services.

graph TD
    A[Lead Sources] -->|Ingest| B(Lead Store / DB)
    B -->|New Leads| C[ICP & Persona Agent]
    C -->|Enriched Data| D[Email Generator Agent]
    K[Knowledge Base / RAG] -->|Retrieve Context| D
    D -->|Drafts| E[Review Queue]
    E -->|Approved| F[Send Orchestrator]
    F -->|SMTP/API| G[Inbox Listener]
    G -->|Replies| H[Reply Classifier Agent]
    H -->|Interested| I[Notify & Metrics]

Loading

📂 Project Structure

/
├── backend/
│   ├── agents/            # Intelligent Agents (Research, Drafting, Classification)
│   ├── api/               # FastAPI Server & Endpoints
│   ├── core/              # Config, LLM Client (OpenAI/Mock)
│   ├── services/          # Business Logic (Ingest, Sender, Risk Control)
│   └── storage/           # SQLite DB & Models
├── frontend/              # Admin Dashboard (served statically by backend)
├── scripts/               # Demo & Utility Scripts
└── docs/                  # Documentation

⚙️ Configuration & Installation

1. Install Dependencies:

   pip install -r requirements.txt

2. Environment Setup: Create a .env file to enable real AI features (optional, defaults to Mock mode).

   OPENAI_API_KEY=sk-...  # Required for real content generation
   SLACK_WEBHOOK_URL=...  # Required for Slack notifications
   MOCK_LLM=False         # Set to True to save costs during dev

3. Run Manually: If you prefer not to use the pilot script:

   # Start Server
   python3 -m backend.api.server
   
   # Visit UI
   # http://localhost:8000

🛠 Tech Stack

• Backend: Python 3.9+, FastAPI, SQLite, ChromaDB (Vector Store)
• AI/LLM: OpenAI GPT-4o (or interchangeable provider via LLMClient)
• Frontend: Vanilla JS/CSS (Lightweight, Dark Mode)
• Deployment: Ready for Docker/Cloud Run (Stateless API + Persistent DB volume)