AI Berkshire
A free toolkit that has an AI assistant research a stock the way famous long-term investors like Warren Buffett would — checking the business fundamentals carefully instead of chasing short-term price moves.
🔗 Visit AI Berkshire
Description
Value investing — buying good businesses at a fair price and holding them for years — takes a lot of careful research most people don't have time to do properly: reading financial statements, understanding competitive advantages, checking management quality. AI Berkshire turns Claude Code into a research assistant that works through that process using frameworks borrowed from well-known value investors, producing structured analysis instead of a quick, shallow take.
AI Berkshire is a free, open-source (MIT license, ~13,900 GitHub stars) Python framework built on Claude Code (and OpenAI Codex) that applies investment methodologies associated with Warren Buffett, Charlie Munger, Duan Yongping and Li Lu, offering 20 specialized investment research skills, a "four-master perspective" framework with adversarial analysis (checking a thesis from multiple angles), multi-agent parallel research coordination, and decimal-precision financial calculations. The project maintains a real, tracked portfolio with documented performance (+69.29% in 2024, +66.38% in 2025 per its own reporting) and is community-driven via a WeChat public account alongside GitHub.
💬 Our review
The short version: if you already understand value investing and want an AI research assistant that applies well-known frameworks rigorously rather than giving generic stock-tip answers, AI Berkshire is a genuinely structured, free tool worth trying.
The self-reported portfolio returns (+69% and +66% across two years) are eye-catching, but treat them the way you'd treat any self-reported trading track record: unaudited, from a small sample of years, and not necessarily replicable given how much market-specific luck plays into short multi-year windows — a framework producing good analysis doesn't guarantee future returns, and past performance is never a promise. The real value here is the structured research process itself (adversarial four-perspective analysis, decimal-precision calculations) rather than the headline return numbers, which should be treated as an interesting data point, not proof the system works. Best suited for someone who already knows how to evaluate an investment thesis and wants an AI-assisted second opinion, not someone looking for automated stock picks to blindly follow.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Entièrement gratuit et open source (licence MIT).
Pros
Gratuit, open source, méthodologie structurée (analyse contradictoire à 4 perspectives)
20 compétences de recherche spécialisées
Portefeuille réel suivi avec performance documentée
Compatible Claude Code et OpenAI Codex
Cons
Performances auto-rapportées, non auditées, sur un échantillon court (2 ans)
Performance passée ne garantit rien pour l'avenir
Nécessite déjà une compréhension de l'investissement value pour en tirer profit
Pas de site web dédié, distribution via GitHub et WeChat uniquement