Onton

Next generation e-commerce search. We're hiring.

Onton is a San Francisco company building a product search engine for consumer goods, and this launch introduces Ontology 1, a neurosymbolic model for complex, conversational, multimodal product search . The pitch is aimed at shoppers who type intent-heavy queries (think "mid-century modern green velvet sofa under $1,200") and at the agentic shopping systems that increasingly stand between people and retailers. According to the launch tweet, the model is at least 2.7x more accurate than existing e-commerce search engines, and Onton positions it against Amazon and Google Shopping on intent-heavy queries . In the demo, co-founder Zach Hudson walks through searches that go beyond keyword and vector retrieval, including image-plus-text prompts ("like this lamp, but more inexpensive"), multi-object screenshots that trigger simultaneous searches across categories, and full mood-board canvases the system parses for vibe, product hints, and constraints at once. The model learns concepts on the fly and caches them, so repeated or related queries run roughly 100x faster after the initial learning pass, and results come with explanations of why each item matched. The timing matters because Onton is pivoting from a consumer search bar toward infrastructure for the broader agentic web. Co-founders Alex Gunnarson and Zach Hudson started the company (formerly known as Deft) after Alex spent 30 hours searching for a specific couch , and the company recently raised $7.5 million in seed funding to expand the work. Ontology 1 is live in Onton's own search today, with an API planned so other builders can plug it into shopping agents and retrieval pipelines outside e-commerce.

Tags

Comments (7)