OpenBusiness
OpenBusiness is an open-source Python CLI for business model research that generates Markdown reports with claims labeled as verified, inferred, or missing. It solves the problem of unstructured and unverified AI-generated company analyses by providing evidence-labeled, reproducible reports. It is built for founders, investors, consultants, product teams, and researchers who need a first-pass company report before deeper diligence.
✨ Key features
- Multi-agent pipeline built with LangGraph
- Evidence collection via Tavily, Firecrawl, and SEC EDGAR
- Explicit evidence labels: verified, inferred, missing
- Domain-specific analysis packs (SaaS, public, ecommerce, etc.)
- Audience-specific report templates (investor memo, founder copycat, etc.)
- Bilingual reports in English and Simplified Chinese
🎯 Use cases
- Founders researching a market to copy, avoid, or attack
- Investors screening a company for durability and risk
- Consultants understanding how a company makes money
- Product teams analyzing customer adoption and switching triggers
- Researchers identifying missing data before making claims
📦 Installation
🧰 Requirements: Python 3.x, API keys for LLM provider (OpenAI, Anthropic, or DeepSeek) and optionally Tavily and Firecrawl.
git clone https://github.com/wanikua/OpenBusiness.git
cd OpenBusiness
./install.sh
The installer asks for language, checks Python, creates a virtual environment, installs the package, and starts the configuration wizard. If using the default .venv, activate it in each new terminal:
source .venv/bin/activate
🚀 Usage
source .venv/bin/activate
openbusiness analyze "Notion" --domain notion.so --language en
The report is written to output/notion_business_model.md.
❓ FAQ
What evidence sources does OpenBusiness use?
It uses Tavily search, Firecrawl scraping, and SEC EDGAR for public companies.
Can I generate reports in languages other than English?
Yes, OpenBusiness supports English and Simplified Chinese via the --language option.
How are claims labeled in the report?
Claims are labeled as [VERIFIED:url], [INFERRED], or [MISSING] to indicate the level of evidence support.
Can I customize the analysis for a specific industry?
Yes, you can use built-in analysis packs like saas or public-company, or provide custom TOML pack files.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.