ResearchPilot AI is a production-grade multi-agent SaaS platform that automates deep research workflows.

Instead of relying on a single prompt-response model, ResearchPilot uses a sequential AI agent pipeline where specialized agents collaborate to:

• Understand user intent
• Build an investigation strategy
• Gather grounded real-time information
• Generate executive-grade research reports

Built with Google Agent Development Kit (ADK) and powered by Gemini 2.5 Flash.

Three autonomous agents collaborate in sequence:

Integrated with Google Search to ensure:

• Fresh information
• Reduced hallucinations
• Source-backed outputs
• Reliable market / trend research

• Firebase Google Authentication
• JWT Token Validation
• Rate Limiting with slowapi
• Protected research endpoints
• Session cleanup workflows

Uses Vertex AI Session + Memory systems for:

• Research continuity
• Context retention
• Multi-session intelligence

Modern responsive UI built using:

• React
• Vite
• Tailwind CSS
• Glassmorphism design system

graph TD

A[User Query] --> B[Planner Agent]
B --> C[Research Plan]

C --> D[Researcher Agent]
D --> E[Google Search]

E --> F[Evidence + Sources]

F --> G[Writer Agent]

G --> H[Final Research Brief]

git clone https://github.com/yourusername/researchpilot-ai.git
cd researchpilot-ai

Create .env

GOOGLE_CLOUD_PROJECT=your-project-id
GOOGLE_CLOUD_LOCATION=us-central1
AGENT_ENGINE_ID=your-engine-id
FIREBASE_PROJECT_ID=your-firebase-id
GOOGLE_APPLICATION_CREDENTIALS=./service-account.json
MODEL_ID=gemini-2.5-flash

Install dependencies:

cd backend

python -m venv venv
source venv/bin/activate

pip install -r requirements.txt
uvicorn main:app --reload --port 8080

Create:

frontend/.env.local

Add:

VITE_FIREBASE_API_KEY=your-key
VITE_FIREBASE_AUTH_DOMAIN=your-domain
VITE_FIREBASE_PROJECT_ID=your-project
VITE_API_URL=http://localhost:8080

Run frontend:

cd frontend
npm install
npm run dev

1. User signs in with Google
2. Firebase issues secure JWT
3. Backend verifies token
4. User UID attached to request
5. Research history saved privately

5 Research Requests / Minute / User

Helps protect:

• API abuse
• Bot traffic
• Excessive token costs

Run end-to-end pipeline tests:

bash test_e2e.sh

researchpilot-ai/
│── backend/
│   ├── agents/
│   ├── middleware/
│   ├── services/
│   └── main.py
│
│── frontend/
│   ├── src/
│   ├── components/
│   └── pages/
│
│── test_e2e.sh
│── README.md

• Market Research
• Startup Validation
• Competitor Analysis
• Academic Summaries
• Industry Reports
• Trend Discovery
• Product Research

• PDF Export Reports
• Team Workspaces
• Citation Viewer
• Streaming Responses
• Voice Research Assistant
• Research Templates
• Slack / Notion Integrations

Contributions are welcome.

1. Fork repository
2. Create feature branch
3. Commit changes
4. Submit Pull Request

Licensed under the Apache 2.0 License