Andreessen Horowitz‘s Marc Andreessen ruffled feathers last week when he claimed that the early-stage VC will be one of the last jobs not done by AI agents.
“There’s an intangibility to it, there’s a taste aspect, the human relationship aspect, the psychology [...] and when the AIs are doing everything else, that may be one of the last remaining fields that people are still doing,” he said.
Putting aside the obvious self-importance and major holes in his argument, is he right?
As AI tools get increasingly sophisticated at assessing markets, modeling financials and sourcing new customers, the value of the human venture investor—especially the junior analyst and principal—is rapidly diminishing.
And these tools are prompting VCs to reconsider whether they really need to hire young talent and to ask: once everything that can be automated is done by software, what is left?
Less is more
“The thing we’re seeing over and over again is you just don’t need to hire in the same capacity anymore,” said Jackie DiMonte, general partner at pre-seed and seed firm Grid Capital. When DiMonte and her partner set up the firm last year, they originally planned to hire junior investment staffers.
“Our hypothesis has completely reoriented to: we want one person who’s a chief of staff, capable of running experiments, launching initiatives—still helpful on the deal process—but much more oriented on how we expand our influence and impact as a firm,” she said.
Many GPs have created custom workflows with OpenAI’s ChatGPT or Anthropic’s Claude, and they are finding AI to be many times faster at market modeling and data collection than an entry-level venture analyst.
Max Ruderman, founder of Harmonic, a startup that runs a natural language processing model for deals sourcing, said he’s increasingly seeing senior VC partners use the software.
VC associate programs have been a crucial on-ramp into a career in the industry. But a growing number of startups are building software tools that automate much of that junior-level work.
In 2023, Tribe Capital incubated and spun out Termina, a quantitative due diligence AI startup founded by Alex Chee, one of Tribe’s earliest employees. Others use Index-backed Hebbia, which offers due diligence and portfolio management tools.
And the sands are not only shifting under VCs: Norway’s sovereign wealth fund said it expects to save $400 million annually by using predictive public markets trading software. A private markets equivalent doesn’t feel too far off.
Picking Michael Jordan in kindergarten
Andreessen argues that no amount of data automation can replace the human je ne sais quoi of early-stage investing. If this is true, the ideal candidate for junior venture jobs will dramatically change over the next few years.
“If discovery [of startups] does get easier, then you have to be good at winning the deal, which means earning the trust of the founder,” said Ruderman.
Applicants from quant shops and hedge funds may well be passed over in favor of applicants with real experience as a founder, incredible charisma or a particularly deep startup network, especially at early-stage firms.
“Early-stage VC is like picking Michael Jordan in kindergarten— when there isn’t much data available to feed into an AI model,” said Keval Desai, managing partner at early-stage firm Shakti.
Overly relying on data-driven, pattern-matching models for investing risks solidifying bias: the founders who are already overlooked would find themselves further cut off.
But despite the risks and ambiguities, early-stage VCs have iterated on purely data-driven approaches investing for a while: SignalFire‘s unapologetically data-driven strategy is a prominent example.
And GV (formerly Google Ventures) infamously used an algorithm that permitted or prohibited new investments informed by a suite of market data before scrapping the approach in 2022.
GV’s “machine” was far less sophisticated than today’s portfolio management software, which will itself likely feel rudimentary in a few years.
“Now maybe there could come along some other architecture different from LLMs that could replace the early-stage VC,” said Desai.
“We will have to wait for that.”
Featured image by Drew Sanders/PitchBook News
Learn about our editorial standards.
