5 startups that caught VCs’ attention at the latest PearX demo day

AI startups are achieving more specialized capabilities, from spatial computing to on-device inference, and this demo day highlights specific advancements beyond broad language models.

PearX, an accelerator run by Pear VC, recently held its demo day in San Francisco, showcasing 16 startups to venture capitalists. The program, known for its selective cohort of 20 startups and substantial investment potential up to $2 million, aims to keep its participants under wraps until demo day, distinguishing it from accelerators like Y Combinator. Among the startups that garnered significant investor attention were Speridlabs, developing spatial foundational models for robotics and gaming with a unique querying capability; Saia, creating a cost-efficient chip for on-device AI inference that bypasses traditional memory requirements; Ren, a secure AI personal assistant emphasizing on-device data and user-set guardrails; Veros, an AI-native trust and estate planner already managing $250 million in assets; and Datum, an AI engineer tool designed to expedite physical product development by indexing and identifying 3D design parts. Investor interest was noted for these five companies due to their innovative approaches to current market challenges.

The emergence of specialized AI hardware and foundational models at PearX signals a maturing AI market, moving beyond general-purpose LLMs towards niche applications. Companies like Saia are directly addressing hardware limitations in AI inference, proposing solutions that challenge established players by rethinking memory architecture and power efficiency. This focus on tangible improvements in speed, cost, and energy consumption is crucial for the wider adoption of AI in edge devices and complex simulations.

The trend of AI being applied to complex, high-value industries like finance and industrial design is evident in Veros and Datum. Veros's AI-driven trust and estate planning, already managing significant assets, indicates a push towards automating professional services that were previously cost-prohibitive or time-consuming. Similarly, Datum's approach to accelerating physical product development through AI-powered design indexing addresses a critical bottleneck in manufacturing and engineering, suggesting AI's expanding role in the physical product lifecycle.