Key Takeaways
- • Former Goldman Sachs machine learning head Charles Elkan and former Capital Group portfolio manager Alex Popa are raising capital for Poesis, a Menlo Park-based AI-driven investment firm, according to Bloomberg.
- • Poesis' inaugural strategy, AI Clarity, has managed the founders' own capital since August and is scheduled to open to external investors in early 2027.
- • AI Clarity invests in 80 to 100 US-listed large-cap stocks with an average holding period of three years, using autonomous AI agents for research and trade decisions under human-set risk parameters.
- • AI Clarity declined 0.3% in October versus a fall of almost 1% for the S&P 500, per people familiar with its performance; the fee structure is 1% management and 20% performance.
- • Poesis, which employs 10 people, plans to expand into a multistrategy hedge fund for institutions and says its AI-led model could eventually manage billions of dollars with fewer than 100 staff.
Summary
Former Goldman Sachs and Capital Group executives have launched Poesis, an AI-driven investment firm based in Menlo Park, California, that will use autonomous agents to research, select and trade equities, according to a report by Bloomberg. The firm is led by Alex Popa, a former portfolio manager at Capital Group, and Charles Elkan, previously head of machine learning at Goldman Sachs. Both are currently raising capital for the firm.
The firm's inaugural strategy, AI Clarity, has been managing the founders' own capital since August and is scheduled to open to external investors in early 2027. The strategy holds a portfolio of 80 to 100 US-listed large-cap stocks with a target average holding period of three years. AI agents use quantitative models to conduct research, identify opportunities and determine trades, while human portfolio managers set risk parameters and approve transactions. The system can also hedge market exposure by shorting equity indices when it identifies potential downturns.
According to people familiar with its performance, AI Clarity declined 0.3% in October, compared with a drop of almost 1% for the S&P 500. The strategy aims to outperform broader equity markets while limiting exposure to major drawdowns. It will charge a 1% management fee and a 20% performance fee.
Poesis currently employs 10 people, including engineers and data scientists. Its founders plan to add strategies and eventually establish a multistrategy hedge fund for institutional investors. The firm's AI-led operating model could allow it to manage billions of dollars with fewer than 100 employees, well below the staffing of established multistrategy managers such as Citadel and Millennium Management.
The launch reflects growing interest in moving AI beyond research and decision support toward more autonomous portfolio management. Poesis has tested fully autonomous AI trading agents in concentrated long-term equity and market-neutral quantitative strategies, with both operating within their prescribed mandates.
The firm's inaugural strategy, AI Clarity, has been managing the founders' own capital since August and is scheduled to open to external investors in early 2027. The strategy holds a portfolio of 80 to 100 US-listed large-cap stocks with a target average holding period of three years. AI agents use quantitative models to conduct research, identify opportunities and determine trades, while human portfolio managers set risk parameters and approve transactions. The system can also hedge market exposure by shorting equity indices when it identifies potential downturns.
According to people familiar with its performance, AI Clarity declined 0.3% in October, compared with a drop of almost 1% for the S&P 500. The strategy aims to outperform broader equity markets while limiting exposure to major drawdowns. It will charge a 1% management fee and a 20% performance fee.
Poesis currently employs 10 people, including engineers and data scientists. Its founders plan to add strategies and eventually establish a multistrategy hedge fund for institutional investors. The firm's AI-led operating model could allow it to manage billions of dollars with fewer than 100 employees, well below the staffing of established multistrategy managers such as Citadel and Millennium Management.
The launch reflects growing interest in moving AI beyond research and decision support toward more autonomous portfolio management. Poesis has tested fully autonomous AI trading agents in concentrated long-term equity and market-neutral quantitative strategies, with both operating within their prescribed mandates.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.