NYTech
People
Personinvestor

Jason Shuman

Jason Shuman is a venture capitalist known for his work as a partner focused on physical AI at Primary Venture Partners, a New York-based seed-stage firm, where he has been active since 2018 [1][5]. He came to venture capital by way of entrepreneurship, having launched a bootstrapped direct-to-consumer footwear company as an undergraduate at the University of Miami in 2011, a venture he ran for about a year in school and a year afterward before winding it down [13][14][18][19]. To break into the industry without a conventional entry point, he drove for Uber at night while sourcing startup deals out of Boston for New York venture firms during the day, leveraging his founder network to build relationships that eventually led to a job offer at Corigin Ventures, where he worked under David Goldberg on proptech, marketplace, and fintech investments [19][22]. At Primary, he went on to lead the firm's consumer investment practice before shifting his focus to physical and vertical AI, and he has built a portfolio that includes lead seed investor status at Dandy and seed or board positions at companies such as Standard Bots, LightTable, Bobyard, Fuse, Ply, and Marker Learning [6][7][8][9][10][11][12].

Shuman describes Primary as the largest seed fund dedicated to New York City, distinguished by a concentrated investing model in which each partner completes only three to four deals annually and specializes deeply within a given market through expert and customer networks [18]. He characterizes the firm's approach as "builders not backers," pointing to a portfolio impact team that grew from roughly 30 to as many as 60 people, supporting portfolio companies with recruiting, go-to-market execution, and strategic finance [17][18]. He also credits Primary Labs, an internal startup studio run by partner Brian Schechter, with generating new venture ideas by identifying business-model patterns that succeeded in one industry and applying them to adjacent ones, then pairing those ideas with vetted founders to increase their conviction before launch [17][16]. According to Shuman, Primary's early funds have seen about 90 percent of portfolio companies raise a Series A and roughly 20 percent reach unicorn status [16].

On investment strategy, Shuman has articulated a thesis centered on vertical AI, arguing that the most valuable companies abstract away complexity and manual human effort in the way Stripe abstracted compliance and fraud detection rather than merely processing transactions [16]. He identifies two viable entry points for vertical AI startups: high-velocity "wedge" products that produce an immediate, visible customer reaction, and AI-native systems of record, which he views as harder to sell because they require a full rip-and-replace of existing infrastructure [16]. He has also expressed a preference for hardware-plus-software approaches, arguing that hardware can capture data invisible to or error-prone for manual human input, feeding higher-quality information into downstream software workflows [16]. Shuman has pointed to Anthropic's Series A, which he notes was passed on by every tier-one venture firm, as evidence that pattern recognition around founders still matters even as the most consequential companies often appear non-consensus at the earliest stages [16].

Earlier in his career, while focused on consumer, fintech, digital health, marketplace, and proptech investing, Shuman emphasized evaluating marketplace businesses for "promiscuity" between supply and demand sides, meaning the degree to which participants transact with different counterparties rather than repeatedly with the same one, as a safeguard against transactions leaking off-platform [19]. He has also framed his broader investing philosophy around democratizing access to tools and services historically reserved for wealthy individuals or large corporations [19].

Insights & ideas

Jason Shuman argues that the winning vertical AI companies succeed by abstracting away complexity and human effort, comparing this to how Stripe's real genius was hiding compliance, fraud, and complexity rather than merely processing payments, a bar he believes vertical AI must clear [1]. He identifies two distinct entry points for building these companies: high-velocity wedge products that create immediate "oh my god" reactions from customers, versus AI-native systems of record, which he considers harder to execute because they require a rip-and-replace motion against entrenched incumbents, leading him to favor backing wedge-first approaches [1].

He is also a strong proponent of combining hardware with software, believing hardware can capture visual and non-visual data that human eyes miss or that people currently input manually and error-prone into systems, unlocking downstream automated workflows [1]. On picking generational companies, he maintains that founder pattern-matching still matters, though he notes the best companies often look non-consensus early on and only become consensus over time, even as rising seed valuations increase the risk of that pattern-matching approach [1].

Experience

  1. Partner
    Primary Venture PartnersSep 2018 to Present
  2. Lead Seed Investor
    DandyApr 2019 to Present
  3. Board Member
    LightTableFeb 2025 to Present
  4. Board Member
    BobyardDec 2023 to Present
  5. Board Member
    FuseApr 2021 to Present
  6. Board Member
    PlyJun 2023 to Present
  7. Seed Investor
    Standard Bots2017 to Present
  8. Board Member
    Marker LearningAug 2021 to Present

Education

Media & appearances

In the news

This page shows public professional information only, each fact cited. Is this you? Corrections or removal within 24 hours, no questions asked.