OpenBusiness is an open-source Python CLI for business model research. Give it a company name, domain, and optional stock ticker; it collects evidence from the web, runs analyst agents, builds a business model canvas, stress-tests assumptions, and labels each claim as verified, inferred, or missing.

Install

pipx install openbusiness
openbusiness config
openbusiness analyze "Notion" --domain notion.so

Or one-shot via uv:

uvx openbusiness analyze "Notion" --domain notion.so

What ships in 0.1.0

• 9-agent linear pipeline (Evidence Collector through Finalizer), built on LangGraph
• Evidence sources: Tavily, Firecrawl, SEC EDGAR
• LLM providers: OpenAI, Anthropic, DeepSeek
• Evidence labels: [VERIFIED:url], [INFERRED], [MISSING] — preserved across every stage
• Bilingual report output (--language en, --language zh)
• 6 domain analysis packs: SaaS, public companies, ecommerce, AI infra, consumer apps, general
• 4 audience templates: standard, investor memo, founder copycat, competitor map
• Unit economics computed in Python (LTV / CAC / breakeven) rather than by the LLM
• Two-tier model routing: smaller models for analyst stages, deep model for synthesis and stress test — typical run 0ドル.10-0ドル.40

Sample reports

The repository ships with three example outputs:

• Notion — PLG SaaS, modeled LTV/CAC 1.6x with churn as the swing assumption
• Vercel — developer platform
• OpenAI — mixed B2C / B2B / API

Links

• PyPI: https://pypi.org/project/openbusiness/0.1.0/
• Docs: https://github.com/wanikua/OpenBusiness#readme
• llms.txt: https://github.com/wanikua/OpenBusiness/blob/master/llms.txt
• License: MIT