AI-First Hedge Fund Launched by Instacart Co-Founder: A Test for Algorithmic Decision-Making, Acc…

In a recent LinkedIn post, Linas Beliūnas discusses the groundbreaking launch of Abundance, a new hedge fund co-founded by Instacart’s Apoorva Mehta, which notably features AI agents as its sole portfolio managers. Beliūnas frames this development as a significant departure from traditional quantitative finance, where human judgment has long been considered paramount.

Beliūnas highlights the core innovation of Abundance: the complete delegation of investment decisions to artificial intelligence. He writes:

Abundance is making the opposite bet: the human is the bottleneck 🤖

According to Linas Beliūnas, the fund’s AI agents are designed to handle the entire investment process, from idea generation and research to stock selection, position sizing, risk management, and execution. This ‘swarm’ of AI agents operates autonomously, with no human portfolio manager making the final call.

The Autonomous Investment Decision-Maker

Linas Beliūnas points out that Abundance is not merely using AI to augment human analysts but is fundamentally replacing them in the decision-making loop. The fund’s structure, as detailed by Beliūnas, involves a team of 10 quants and engineers who build and maintain the system, while the AI agents themselves have been trading proprietary capital for nine months, with Mehta claiming outperformance against several benchmarks.

Beliūnas notes the significant backing Abundance has received, including $100 million in seed equity. The potential expansion into outside capital, bonds, commodities, and derivatives further underscores the ambitious nature of this AI-driven venture.

The real test will be when markets break, liquidity vanishes, correlations spike, and every agent suddenly “reasons” toward the same exit.

Challenges and the ‘Turing Test’ for AI Funds

Despite the initial success claimed by Abundance, Linas Beliūnas injects a note of caution, emphasizing that crucial data points are still missing. He lists several caveats:

• No audited numbers yet.
• No named benchmarks.
• No real drawdown test.

Beliūnas references Citadel Securities CEO Ken Griffin’s previous assertion that generative AI had not yet helped hedge funds consistently beat the market. Abundance aims to challenge this perspective.

A central theme in Beliūnas’s analysis is the potential vulnerability of AI-driven strategies in volatile market conditions. He argues that the true test for Abundance, and potentially for other AI-first funds, will come during periods of market stress.

The Risk of Synchronized Behavior

Linas Beliūnas raises a critical concern about diversification in an era of increasingly sophisticated AI models. He posits:

Because if dozens of AI funds use similar models, data, and tools, diversification may start to look a lot like synchronized crowd behavior.

In Beliūnas’s view, if multiple AI funds employ comparable methodologies, their reactions to market shocks could become highly correlated, leading to a ‘herd’ effect that could exacerbate losses. This scenario, he suggests, represents the first true ‘Turing test’ for AI in money management—assessing whether these systems can navigate unpredictable market turmoil with genuine intelligence rather than simply mimicking collective behavior.