Skip to main content
This Week in Startups

Neurosymbolic AI outperforms chatbots and product search | E2327

54 min episode · 2 min read
·
Zach Hutson

Episode

54 min

Read time

2 min

Topics

Startups, Fundraising & VC, Design & UX

AI-Generated Summary

Key Takeaways

  • Neurosymbolic vs. LLM search: For ecommerce discovery above $50 where taste and self-expression drive purchases, LLMs fail by regressing to statistically probable results. Onton's Ontology One model updates its understanding in real time without retraining, outperforms major search engines by 2.5x, and costs 1/1000th of a standard frontier model training run in the US.
  • Real-time model learning for ecommerce: Unlike transformer-based models requiring full retraining cycles to incorporate new data, neurosymbolic architectures can update their graph database continuously. Onton's model learns product attributes like material, texture, and aesthetic style from each search, meaning accuracy compounds with every query without additional compute expenditure or engineering intervention.
  • API-first expansion strategy: Onton is releasing Ontology One as an API rather than a Shopify plugin because the model generalizes beyond home decor into apparel, electronics, and non-ecommerce domains. Cross-category learning accelerates expansion — knowledge of polyester in furniture automatically improves apparel search — reducing the labeling and setup costs historically required when entering each new product vertical.
  • User-generated content as a competitive moat: As LLMs cannibalize blog posts, reviews, and editorial content by absorbing and replacing them, ecommerce platforms without native user-generated content lose training signal. Building mood boards, canvases, and shareable user collections directly into the product creates proprietary behavioral data that external AI cannot easily replicate or scrape away.
  • In-space refueling demand is immediate, not speculative: Spacium has secured close to $100M in commercial contracts and over $2B in LOIs for orbital refueling before their full station is operational. Spacecraft designers are already building future missions around the assumption of refueling availability, meaning the market exists now — the bottleneck is deploying the physical infrastructure, not generating customer demand.

What It Covers

Two startup founders present contrasting frontier technologies: Zach Hutson of Onton demonstrates a neurosymbolic AI search engine delivering 2.5x better ecommerce results than major search companies at 1/1000th the training cost, while Aashi Dissanayake of Spacium explains in-orbit fuel transfer technology backed by $2B+ in letters of intent.

Key Questions Answered

  • Neurosymbolic vs. LLM search: For ecommerce discovery above $50 where taste and self-expression drive purchases, LLMs fail by regressing to statistically probable results. Onton's Ontology One model updates its understanding in real time without retraining, outperforms major search engines by 2.5x, and costs 1/1000th of a standard frontier model training run in the US.
  • Real-time model learning for ecommerce: Unlike transformer-based models requiring full retraining cycles to incorporate new data, neurosymbolic architectures can update their graph database continuously. Onton's model learns product attributes like material, texture, and aesthetic style from each search, meaning accuracy compounds with every query without additional compute expenditure or engineering intervention.
  • API-first expansion strategy: Onton is releasing Ontology One as an API rather than a Shopify plugin because the model generalizes beyond home decor into apparel, electronics, and non-ecommerce domains. Cross-category learning accelerates expansion — knowledge of polyester in furniture automatically improves apparel search — reducing the labeling and setup costs historically required when entering each new product vertical.
  • User-generated content as a competitive moat: As LLMs cannibalize blog posts, reviews, and editorial content by absorbing and replacing them, ecommerce platforms without native user-generated content lose training signal. Building mood boards, canvases, and shareable user collections directly into the product creates proprietary behavioral data that external AI cannot easily replicate or scrape away.
  • In-space refueling demand is immediate, not speculative: Spacium has secured close to $100M in commercial contracts and over $2B in LOIs for orbital refueling before their full station is operational. Spacecraft designers are already building future missions around the assumption of refueling availability, meaning the market exists now — the bottleneck is deploying the physical infrastructure, not generating customer demand.

Notable Moment

Spacium's two-person founding team built and launched their first payload in five months entirely in-house, achieving the highest-precision robotic actuator ever tested in orbit. An early SpaceX advisor grew so concerned by their silence during the build that he feared something had happened to them.

Know someone who'd find this useful?

Episode Transcript

LLMs are not good at this. They regress to the mean and they find the most probable thing and that's often not what you need personally. When you need to remove hallucination and you need trust, you need to be able to see inside the model, that's where neurosymbolic models can really shine. You really have built something between Houzz and Pinterest that is more powerful than both of them. The performance of it is already 2.5 x greater than some of the largest search companies in the world. It learns that new information, so you don't have to go do another training run. It's literally updating in real time. It's one one thousandth the cost of an average frontier model training run-in The US. This Week in Startups is brought to you by wisecurity, the on demand security team for startups. Need enterprise grade security without hiring a $400,000 c I s o? Security gives you 40 plus expert engineers matched to exactly what you need by the hour with your first six hours completely free. Go to ysecurity.io/twist.agree.com. Stop chasing invoices and automate your entire contract to cash stack. Go to agree.com and tell them Jason sent you to get 50% off for life. And Vanta. Compliance and security shouldn't be a deal breaker for startups to win new business. Vanta makes it easy for companies to get a SOC two report fast. Get $1,000 off for a limited time at www.vanta.com/twist. We got Zach Hutson joining us. He's the CEO of Onton. They're a search and discovery engine for ecommerce. It's powered, Jason, by a neurosymbolic AI model. What does that mean? You could see it for yourself at onton.com. We're gonna find out right now. Zach, thank you so much for joining us. Thanks for having me. Pleasure. Yeah. And you can give us a demo. So let's let's, you know, show, don't tell. Yeah. Oh, yeah. Yeah. Yeah. What have you built? Yeah. We can dive right in. Let's do it. Oh, and we're investors in the company, just as a full disclosure here. Yeah. I'll give your audience a chance to go and do some some basic searches for themselves. But why don't I jump in and do some, more of the mind blowing searches, and a little bit of a little bit of background about Anton. We have these things called surfaces within our product. If you've ever you have have Lon or Jason, have either of you used, Notion before? I'm sure you have. Plenty of time. Yeah. More than once. So Yeah. They have these things called blocks there. They let you organize information in different ways. People do this quite often in shopping journeys too, where they create lists, they create mood boards, these types of things. These are surfaces within Anton. And so what I've prepared for you, was I put together, some of your favorite things, Jason, or at least some of the hotels that you've mentioned, some …

Get the full transcript (10,550 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all This Week in Startups transcripts →

You just read a 3-minute summary of a 51-minute episode.

Get This Week in Startups summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

Books, tools, and gear mentioned in this episode

SignalCast may earn commission on purchases via these links.

Tools

  • Ontology OneBy guest

    by Onton

    Onton's Ontology One model updates its understanding in real time without retraining, outperforms major search engines by 2.5x, and costs 1/1000th of a standard frontier model training run in the US.

company

  • SPONSORS: Vanta at https://vanta.com/twist
  • SPONSORS: Agree at https://agree.com
  • SPONSORS: Y Security at https://ysecurity.io/twist

More from This Week in Startups

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

Explore Related Topics

This podcast is featured in Best Startup Podcasts (2026) — ranked and reviewed with AI summaries.

Read this week's Startups & Product Podcast Insights — cross-podcast analysis updated weekly.

You're clearly into This Week in Startups.

Every Monday, we deliver AI summaries of the latest episodes from This Week in Startups and 192+ other podcasts. Free for one show.

Start My Monday Digest

No credit card · Unsubscribe anytime