🧪 Paper2SaaS

Transform academic arXiv papers into validated SaaS business opportunities using a multi-agent AI factory.

---

📖 Table of Contents

• Overview
• Key Features
• Architecture
• Methodology
• Setup
• Usage
• Python Package Details
• Project Principles

---

🎯 Overview

Paper2SaaS bridges the gap between theoretical academic research and commercial application. By leveraging a team of specialized AI agents, the platform automates the discovery, ideation, and validation process required to turn complex research papers into actionable business plans.

---

✨ Key Features

• 🕵️ Intelligent Paper Analysis: Deep parsing of arXiv papers with multi-tool fallback (Arxiv → Semantic Scholar).
• 💡 Automated Brainstorming: Generates unique SaaS product concepts based on core research innovations.
• ✅ Rigorous Validation: Real-time market research, competitor analysis, and patent risk assessment.
• 📋 Technical Roadmap: Detailed implementation plans, including suggested GitHub repos and MVP timelines.
• 🛠️ Anti-Hallucination Protocol: Enforced structured outputs and multi-step verification (Chain-of-Note, Chain-of-Verification).
• 🖥️ Reactive UI: Modern Next.js interface with real-time streaming of agent reasoning and Claude-style artifacts.

---

🏗️ Architecture

Paper2SaaS employs an orchestration pattern with a primary execution team and a secondary critique team to ensure high-quality, verified results.

Multi-Agent Workflow

graph TD
    User((User)) -->|Input: ArXiv ID| PA[Paper Analyzer]
    
    subgraph "Execution Team (paper2saas_team)"
        PA --> MR[Market Researcher]
        MR --> IG[Idea Generator]
        IG --> VR[Validation Researcher]
        IG --> PE[Product Engineer]
        VR --> SA[Strategic Advisor]
        PE --> SA
        SA --> FC[Fact Checker]
        FC --> RG[Report Generator]
    end
    
    subgraph "Critique Team (idea_roaster_team)"
        IG -.-> DA[Devils Advocate]
        IG -.-> MS[Market Skeptic]
        DA -.-> SA
        MS -.-> SA
    end
    
    RG -->|Output: Opportunity Report| Final((Opportunity Report))

Loading

Specialized Agents

Agent	Role	Focus
Paper Analyzer	Discovery	Fetches and summarizes core research innovations.
Market Researcher	Analysis	Identifies market gaps and target audience pain points.
Idea Generator	Ideation	Synthesizes research and market needs into SaaS concepts.
Validation Researcher	Evidence	Searches for competitors, funding, and real-world demand.
Product Engineer	Engineering	Scopes the MVP and technical stack.
Strategic Advisor	Evaluation	Scores ideas based on feasibility and market potential.
Devils Advocate	Critique	Challenges technical assumptions with verified evidence.
Fact Checker	Verification	Final audit of all claims against tool sources.

---

🧪 Methodology

Anti-Hallucination Measures

To ensure reliability, all agents follow a strict protocol:

1. Structured Outputs: Forced Pydantic schemas for all data.
2. Chain-of-Note (CoN): Agents must take structured notes from search results before synthesizing.
3. Chain-of-Verification (CoVe): A dedicated fact-checking pass verifies every claim against source material.
4. Source Attribution: Every fact in the final report is linked to its source URL or paper citation.

---

🚀 Setup

Prerequisites

• Python: 3.12 or higher
• Node/Bun: For the frontend
• uv: Recommended Python package manager

1. Clone & Install

git clone https://github.com/Ash-Blanc/paper2saas
cd paper2saas

# Install Backend
uv sync

# Install Frontend
cd agent-ui
bun install
cd ..

2. Configuration

The backend looks for a .env file in paper2saas/paper2saas/.env.

cp paper2saas/paper2saas/.env.example paper2saas/paper2saas/.env
# Edit the file with your API keys:
# MISTRAL_API_KEY=your_key
# FIRECRAWL_API_KEY=your_key

---

🔌 Usage

1. Start the Backend (API)

The backend runs on port 7777 by default.

cd paper2saas
uv run python server.py

2. Start the Frontend (UI)

The UI provides an interactive portal to chat with your agents.

cd agent-ui
bun dev

Navigate to http://localhost:3000 to start transforming papers!

---

📦 Python Package Details

Package Structure

src/paper2saas/
├── __init__.py          # Public API exports
├── config.py            # Centralized settings (pydantic-settings)
├── models/              # Shared Pydantic models
├── tools/               # Agno toolkits (Semantic Scholar, etc.)
├── analysis/            # Analysis engines (Citation Graph)
├── agents/              # Agno agent definitions
└── workflows/           # End-to-end pipelines

Library Usage

You can use the core components as a Python library:

from paper2saas import IdeaToSaaSWorkflow

workflow = IdeaToSaaSWorkflow()
result = await workflow.run(seed_paper_id="arXiv:1706.03762")
print(f"Generated {len(result.saas_concepts)} SaaS concepts.")

---

📜 Project Principles

• KISS (Keep It Simple, Stupid): We prioritize clear, maintainable code over complex abstractions.
• Agentic Rigor: Agents are tools, but verification is mandatory. We trust but verify every output.
• Open Discovery: Bridging the gap between ivory tower research and the startup garage.

---

🤝 Contributing

Currently in active development. Main contributor: ash_blanc

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.