Hedge funds may soon deploy fleets of artificial intelligence agents to dramatically expand research capacity and stock coverage, according to a report by Bloomberg citing Divya Nettimi, founder of New York-based hedge fund Avala Global.

Speaking at the Bloomberg Invest conference in New York, Nettimi said that within three to five years, analysts could be supported by AI bots capable of monitoring data flows across hundreds of stocks, filtering signal from noise and surfacing actionable insights for human decision-makers.

“I could see a world where an analyst who previously covered 20 stocks could cover 200, because they are managing a fleet of agents that tracks and analyses them,” Nettimi said.

The productivity gains, she argued, would not be incremental but transformative. AI agents could help funds evaluate a far broader universe of potential investments — expanding the “top of funnel” — before narrowing down to a concentrated portfolio of 15 to 20 high-conviction positions each year.

“The increase is not going to be marginal; it’s going to be an order of magnitude,” Nettimi said, adding that entirely new workflows are likely to emerge as AI tools become embedded in investment processes.

Avala, which manages approximately $2bn, has already developed a firm-wide proprietary AI model that serves as a core input into its investment framework. Nettimi’s former employer, Viking Global Investors, has similarly built an internal chatbot, VikingGPT, to assist portfolio managers in testing and refining trade ideas.

Other managers are exploring comparable applications. At Lone Pine Capital, co-chief investment officer Kelly Granat said AI could help “commoditise the base layer” of fundamental analysis, reducing siloed sector research and creating a more standardised investment vocabulary across teams. Lone Pine manages around $19.5bn in long-short equity strategies.

Despite the enthusiasm, both investors stressed that human judgement remains central to the investment process. Interpreting management teams, asking the right questions and synthesising quantitative and qualitative information – as well as pattern recognition built over decades – are capabilities that Nettimi believes will be harder to automate.

You can read more on this from HedgeWeek, here.