← Home
This Week in Startups · August 19, 2026 · 54m 19s

Neurosymbolic AI outperforms chatbots and product search | E2327

Jason Calacanis and Lon Harris host Zach Hudson of Onton to discuss neurosymbolic AI and how it improves product search by learning individual taste and aesthetics rather than relying on keyword matching. Hudson explains why frontier labs aren't pursuing this approach and how Ontology 1 avoids the black-box problem of large language models. In the second half, Ashi Dissanayake of Spacium joins to argue that following a record-breaking SpaceX IPO, the true bottleneck in the space industry has shifted from launch to orbital refueling infrastructure.

This summary was generated from show notes and public descriptions, not from a full transcript review. Details may contain inaccuracies.

Highlights

•
Neurosymbolic Models Are Not Black Boxes00:00:00
Zach Hudson explains that unlike large language models, which operate as black boxes, Onton's neurosymbolic model combines neural networks with symbolic reasoning, making its decision process transparent and auditable.
•
Personalized Taste Learning Over Keyword Search00:00:00
Onton's model learns a user's taste and aesthetic preferences over time, producing product search results tailored to individual style rather than relying on generic keywords and tags.
•
Why Frontier Labs Aren't Building Neurosymbolic AI00:00:00
Hudson discusses why major AI labs focus on scaling neural networks rather than pursuing neurosymbolic approaches.
•
Orbital Refueling Is Now the Space Bottleneck00:28:13
Ashi Dissanayake of Spacium argues that with launch costs dropping and a record SpaceX IPO, the real constraint in space is no longer getting off the ground—it's refueling spacecraft once they're in orbit.

Editorial

•
The hosts and Hudson discuss how AI-generated content is flooding the internet, and why user-generated content from real human taste-makers becomes a defensive moat for companies that collect it.

Misc

✧Jason Calacanis is a fan of Frank Lloyd Wright's Ennis House and mentions the set design of 'The Studio' was inspired by it.
✧The show notes link to a Startup Archive piece on Paul Graham's advice to Sam Altman: 'always make an API.'
Was this useful?