Onton Launches Ontology 1: Neurosymbolic Search Model Outperforms E-Commerce Giants by 2.7x

San Francisco-based search company Onton has released Ontology 1, a neurosymbolic model for complex, conversational multimodal product search. In independent benchmarks with 90 queries evaluated by three AI judges, Ontology 1 achieved mean precision of 0.630, versus 0.543 for Google Shopping and 0.469 for Amazon, while indexing roughly 1% of their catalogs. The model's strength lies in how it interprets purchase intent. Rather than relying on seller labels that may be absent or misleading, Ontology 1 reasons from objective product properties—fiber type, weave pattern, construction method—to answer complex queries like "pet-friendly sectional." When encountering unknown attributes, the system identifies these as knowledge gaps and constructs answers through explicit, reusable reasoning. Ontology 1 is now live for users at Onton.com, with partner access granted on a case-by-case basis. No public API, pricing tier, or open model checkpoint exists, positioning adoption as a partnership rather than a plug-and-play integration.