AI Berkshire is an AI-driven value investment research framework that integrates methodologies from four investment masters and employs multi-agent parallel analysis.
Source: Official README and linked documentation View on GitHub →AI Berkshire is being discussed because its unique approach of combining human investment wisdom with AI technology. It addresses the pain point of traditional investment research by providing a structured, multi-perspective analysis that enhances decision-making discipline. The project differs through its use of Claude Code and multi-agent parallel analysis, offering a broad and data-driven investment research experience.
Source: Official README and linked documentationFour AI Agents independently research and analyze a company from different perspectives (e.g., business model, financial valuation, industry competition, risk management), providing a broad and multi-perspective analysis.
Source: Official README and linked documentationBased on the methodologies of Warren Buffett, Charlie Munger, and others, the framework provides a systematic and structured approach to investment research.
Source: Official README and linked documentationThe framework incorporates data-richness ratings, cross-verification of key data, and anti-consensus checks to ensure the accuracy and reliability of investment decisions.
Source: Official README and linked documentationThe architecture of AI Berkshire is inferred to be a three-tiered design with a Skill layer, an Agent layer, and a Tool layer. The Skill layer provides 16 distinct entry points for various investment research tasks. The Agent layer consists of four independent Agents that work in parallel to analyze a company from different perspectives. The Tool layer ensures data accuracy and report integrity.
Source: Official README and linked documentationdecimal.Decimalfinancial_rigor.pyAI Berkshire is suitable for investors, financial analysts, and individuals interested in value investing. It is useful in scenarios such as broad company analysis, industry research, portfolio management, and investment decision-making.
Source: Official README and linked documentationAs of 2026-09-23, v1.0.0 is the latest and only formal release. It was published on 2026-04-07 and listed five Claude Code skills with manual-copy installation. Current main lists 20 skills and supplies installation scripts for Claude Code and Codex, so adopters should state whether they follow the tag or main.
Source: github.com/xbtlin/ai-berkshire/releases/tag/v1.0.0status: docs-only | checked_at: 2026-09-23 | environment: Docs only. Reviewed the official README, directory layout, installation-script notes, and GitHub Releases. | limitations: The skills were not installed. No model or financial_rigor.py run was performed. Repository-reported data and the maintainer's published account-return screenshots were not independently checked. | observed_output: The current README lists 20 skills and installation scripts for Claude Code and Codex. The latest and only formal release is v1.0.0; its release page listed five skills at publication time.
Source: github.com/xbtlin/ai-berkshire/releases/tag/v1.0.0FinRobot is a runnable financial-agent platform with a Python package, data sources, valuation modules, and tutorials; it also requires model and financial-data API configuration. AI Berkshire is closer to a set of research rules and skills installed into Claude Code or Codex. FinRobot fits programmatic data pipelines, valuation code, and an application framework. AI Berkshire applies a fixed research process to an existing coding agent. Output from either project should not be used directly as a trading instruction.
Source: github.com/AI4Finance-Foundation/FinRobotAI Berkshire is a project for those interested in integrating AI into value investing. Its unique approach to investment research, combined with the power of Claude Code, offers a tool for enhancing investment decision-making. It is particularly suited for investors and analysts looking for a structured and data-driven approach to investment research.
The current README lists 20 skills and installation scripts for Claude Code and Codex. The latest and only formal release is v1.0.0; its release page listed five skills at publication time. The skills were not installed. No model or financial_rigor.py run was performed.
The review environment was: Docs only. Reviewed the official README, directory layout, installation-script notes, and GitHub Releases. ai-berkshire behavior outside that scope still needs a local check.