Onix's 'Personal Intelligence' Promises Privacy, But Where is the Clinical Rigor?

Montréal-based startup Onix is positioning itself as the antithesis to the data-hungry models of Big Tech, as BetaKit first reported. Founded in 2025 by David Bennahum and Nicholas Nadeau, the company recently announced a $5 million USD ($6.9 million CAD) pre-seed convertible note round led by Los Angeles-based AlphaEdison, with participation from New York-based Ride Home Fund, Kitchener-Waterloo’s Garage Capital, Real Ventures co-founder Jean-Sébastien Cournoyer, and United Talent Agency co-founder Jeremy Zimmer.

Onix offers what it calls "personal intelligence": a library of AI versions of health and wellness practitioners, referred to as "onixes." Users pay for always-on access to these chatbots, which range from holistic nutritionists and former elite athletes to psychiatrists. According to reporting from BetaKit, the company does not train its models on vast streams of user data. Instead, it partners with wellness content creators who provide permission for Onix to train chatbots on their specific work and advice.

From a technical standpoint, the company is betting on efficiency and privacy. According to Bennahum, the cost of inference for Onix is just "one two hundredths" that of a frontier model. To address consumer distrust—citing a 2025 poll where 83 percent of Canadians expressed worries about AI and privacy—Onix runs small models locally on users' devices. This architecture ensures no conversation logs are kept and no data returns to a data center.

However, for those of us wary of "wellness" marketing, the critical question is the nature of the training data. Onix does not rely on peer-reviewed clinical datasets; rather, it trains its AI on the work of practitioners who are invited by the company. This includes unpublished works and public appearances, such as podcast recordings. While the library includes credentialed professionals like family nurse practitioner Ashley Shrader and psychiatrist and author David Rabin, the company admits that the credentials of its "experts" vary.

This creates a potential tension between the app's marketing and its utility. While Nadeau explicitly stated to BetaKit that Onix does not provide medical advice and has guardrails to prevent the AI from straying into that territory, the platform is built on the perceived authority of its contributors. These practitioners, whose fields span from longevity to addiction recovery, set their own subscription fees and receive 70 percent of the resulting revenue, while Onix takes 30 percent.

By shifting the data burden from the user to the "expert," Onix solves a privacy problem, but it introduces a quality-control variable. The value of an "onix" is entirely dependent on the quality of the content provided by the individual creator. Without a standardized clinical benchmark, the platform functions less as a medical tool and more as a scalable distribution system for the personal brands of wellness influencers.