Vision and positioning¶
Fairmark is the valuation engine for private companies and real assets that AI agents can use and analysts can verify.
The problem¶
LLMs are now part of analysts' daily work: they read financial statements, draft memos and build models. They are not reliable calculators, and in a valuation a small arithmetic or convention error changes the answer. Existing open-source tools for agents mostly value listed companies from a ticker, under US/IFRS assumptions.
What makes Fairmark different¶
| Difference | How a user sees it |
|---|---|
| Private companies first | Quickstart starts from financial statements or a business plan, not a ticker |
| Auditable | Every valuation produces a methodology note with formulas, inputs and steps |
| Analyst trust | Results can be re-checked step by step; Excel export with live formulas is planned |
| Determinism | Input hash and verify reproduce any result to the last digit |
| Useful to agents | Function calling, MCP, CLI and skills from one engine; public benchmark planned |
| Everywhere analysts work | Python, CLI, MCP, n8n (planned), browser playground (planned) |
Principles¶
See the constitution.