ai-Geek

Worth watching cautiously, but the evidence is far too limited to call it proven or mature. The current evidence is mainly a single X post describing it as an “AI Berkshire”-style framework. That supports attention and positioning, but it is only heat proof, not usability proof.

By positioning, this is not a brokerage tool, market terminal, or auto-trading bot. It looks more like an open-source multi-agent research workflow built on Claude Code, where value-investing methods are broken into parallel research roles or steps. A better analogy is “a multi-agent workflow project for investment research,” not a finished product that replaces a professional research team.

On setup difficulty and cost, the evidence only supports that it uses a Claude Code multi-agent approach. There is no official pricing, benchmark, or hands-on report in the provided sources, so API cost, runtime cost, and deployment complexity should be treated conservatively as unknown. It likely is not zero-friction for retail users and is more suitable for people comfortable with code, prompting, and research process design.