GrowthPilot
AI-powered B2B outbound sales automation pipeline exposed as an MCP server that transforms natural language goals into qualified sales intelligence and personalized emails.
README
๐ GrowthPilot โ Autonomous Customer Acquisition Platform
GrowthPilot is an AI-powered B2B outbound sales automation pipeline, exposed as an MCP (Model Context Protocol) server coupled with a premium, real-time reactive web dashboard.
By orchestrating multiple external APIs through a robust 7-stage self-correcting workflow, GrowthPilot transforms a simple goal like "Find SaaS companies in Bangalore with 20-100 employees" into qualified sales intelligence and personalized, audit-verified emails ready for outreach.
๐ Live Production URLs
- ๐ป Interactive Frontend (Vercel): https://ui-sigma-snowy.vercel.app
- โ๏ธ Production MCP Backend (NitroStack Cloud): https://growthpilot-6a5af3-hyd-bros-amrita-university-amritapuri-campus.app.nitrocloud.ai
- ๐ Live SSE Handshake Endpoint:
https://growthpilot-6a5af3-hyd-bros-amrita-university-amritapuri-campus.app.nitrocloud.ai/sse
โก Quick Start for Judges (No Installation Required)
You do not need to install or run anything locally to test GrowthPilot:
- Open the Interactive Frontend URL.
- Enter a natural language goal in the input field, for example:
"Find commercial printing businesses in Kerala interested in workflow automation"
- Click Run Campaign.
- Watch the multi-agent pipeline execute in real time across all 7 stages on the live NitroStack Cloud container.
- Once complete (~40 seconds), scroll down to see the leads. Click any card to expand the profile drawer to view the generated email copy and the quality audit score!
๐ง Why MCP? (The Composable Agent Difference)
Standard LLMs with "Deep Research" capabilities can browse the web and answer questions. GrowthPilot takes this further:
- Task Accomplishment, Not Just Chat: Instead of conversational answers, it outputs structured, actionable business deliverables (validated leads + ready-to-send copy).
- Self-Correcting Pipeline: A dedicated Critic Agent audits the output of the Drafting Agent, scoring it and automatically sending it back for revisions if it falls below our quality threshold.
- Open Standard Composability: Because it's an MCP server, any compatible AI client (Claude Desktop, Cursor, ChatGPT, custom CLI tools) can plug directly into the GrowthPilot server and gain these B2B prospecting tools instantly.
๐๏ธ Architecture
graph TD
A["๐ฏ User Goal Input"] --> B["๐ 1. Planner Agent (Gemini)"]
B --> C["๐ 2. Discovery Agent (Google Places API)"]
C --> D["โ
3. Validator Agent (Rule-based Filter)"]
D --> E["๐ฌ 4. Research Agent (Tavily Search API)"]
E --> F["๐ 5. Qualification Agent (Gemini)"]
F --> G["โ๏ธ 6. Drafting Agent (Gemini)"]
G --> H["๐งช 7. Critic Agent (Gemini Audit)"]
H -->|"Score < 0.8 (Up to 3 Revisions)"| G
H -->|"Score >= 0.8"| I["๐ Ready Outbound Campaign"]
The 7-Stage Pipeline
- Planner: Parses user goals into structured campaign parameters (target industry, locations, filters).
- Discovery: Queries Google Places live for matching businesses.
- Validator: A zero-cost deterministic gate that filters out invalid industries, categories, or keywords.
- Research: Crawls the web and scrapes online presence using Tavily.
- Qualification: LLM-based scoring that assigns leads to High, Medium, or Borderline tiers based on fit.
- Draft: Generates hyper-personalized cold outreach emails.
- Critic: Audits emails for tone, placeholders, and length, revising automatically up to 3 times to hit a
0.8quality score.
๐ ๏ธ MCP Primitives Implemented
GrowthPilot provides a complete set of Model Context Protocol features:
- Tools:
gp_run_pipeline: Orchestrates and runs the entire 7-stage campaign end-to-end.
- Resources:
growthpilot://system/status: Exposes real-time server health and active providers.
- Prompts:
gp_campaign_brainstorm: Assists users in brainstorming high-yield campaign goals based on their industry.
๐ป Local Development Setup
The repository is structured as a monorepo containing:
/(Root): The TypeScript + Node.js NitroStack MCP Server./ui: The React + Vite + TypeScript Frontend dashboard.
1. Backend Setup
- Clone the repository and install dependencies:
npm install - Create a
.envfile in the root based on.env.example:PORT=3000 GEMINI_API_KEY=your_gemini_key TAVILY_API_KEY=your_tavily_key GOOGLE_MAPS_API_KEY=your_google_maps_key LIVE_MODE=true - Build and start the development server:
npm run dev - (Optional) Run the test suite:
npm run test
2. Frontend Setup
- Navigate to the
/uidirectory and install dependencies:cd ui npm install - Start the Vite development server (configured to proxy
/sseand/mcpto the backend on port 3000):npm run dev - Open
http://localhost:5173in your browser.
๐ก๏ธ Security & Environment Credentials
- Credentials for all live APIs (Gemini, Google Maps, Tavily) are securely handled at the environment level inside the NitroStack Cloud container. No raw keys are exposed to the browser or checked into source control.
- Production builds compile cleanly with TypeScript ESM imports.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.