Onix Launches Tiny Labs to Build Small Language Models Around Individual Expertise
MONTREAL, Sept. 28, 2026
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.

![]()
Onix Launches Tiny Labs to Build Small Language Models Around Individual Expertise
PR Newswire
MONTREAL, Sept. 28, 2026
MONTREAL, Sept. 28, 2026 /PRNewswire/ — Onix, the Montreal company pioneering Personal Intelligence®, an approach to AI built on the licensed knowledge of individual, named experts instead of scraped internet data, today announced the launch of Onix Tiny Labs, a dedicated research and development group conducting foundational research into small language models designed around individual experts to preserve their knowledge, judgment and voice.
As the world’s largest AI companies race to introduce personalization and creator tools, Onix is investing in a different approach by challenging that bigger isn’t always better. After developing personalized AI experiences with more than 30 health and wellness experts while pushing the limits of off-the-shelf small models, Tiny Labs was created to conduct the foundational research needed to develop a new class of highly customizable small models designed from the ground up around an individual’s knowledge, judgment and voice.
Tiny Labs’ research agenda starts with a measurable problem: models trained on everything converge on averaged answers, while an expert’s voice and judgment established through prompting can drift within just a few conversational turns. Standard customization methods, including fine-tuning and adapters, can fail to preserve an expert’s knowledge or increase hallucination. The lab will research new model architectures, training methods and evaluations designed specifically to preserve the expert’s authenticity.
“Creators and experts are being asked to hand the most valuable parts of themselves, their intellectual property, their voice and their judgment, to massive AI systems that were never designed to protect something that personal,” said David Bennahum, co-founder and CEO of Onix. “Tiny Labs challenges the common wisdom that bigger is always better in language modeling. We believe AI should amplify human intelligence, not average it out or replace it. That has been the founding idea behind Onix from the beginning, and Tiny Labs is our commitment to ensuring that as AI evolves, experts can scale their knowledge without sacrificing what makes it distinctly theirs.”
Based in Montreal, Tiny Labs will advance Onix’s approach to privacy and expert ownership, building within Canada’s strong privacy framework with the goal of creating models that remain distinctly tied to the people whose expertise powers them. These models are small enough to run privately on a user’s phone, without relying on a remote server or centralized data center.
“There’s a technical difference between giving a large model a creator’s content and building a model around that individual from the ground up,” said Nicholas Nadeau, PhD, co-founder and CTO of Onix. “A new interface, memory layer or set of personalization features can make a foundation model feel individual, but the underlying architecture is still designed to aggregate intelligence at massive scale.”
For the experts whose knowledge powers these models, that distinction is critical. For example, nutrition is full of answers that can be technically correct and still be wrong for the person asking the question.
“My expertise isn’t just what I know, it’s how I evaluate someone’s individual needs and determine what matters for them,” said Ashley Koff, RD, Onix weight health expert. “If an AI only repeats what I’ve written online but can’t represent my judgment, it isn’t really giving someone my personalized guidance. Onix is working to preserve my expertise and judgement while protecting their audience’s privacy.”
Onix experts retain ownership and control over how their knowledge is used and receive 70% of the revenue associated with their models. By pairing that economic model with the work of Tiny Labs, Onix aims to give experts a way to scale their knowledge without giving up control of what makes it uniquely theirs.
Onix has built AI models with experts across health and wellness, including William Li, MD; Jordan Metzl, MD; Dave Rabin, MD, PhD; Kwadwo Kyeremanteng, MD; Elissa Epel, PhD; and Ashley Koff, RD. Their expertise spans medicine, nutrition, mental health, longevity and other areas of health and wellness, forming the foundation of Onix’s growing expert ecosystem.
Onix is currently available in early access on iOS. Full commercial availability is coming shortly, with Android to follow.
To learn more, visit Onix.life and follow Onix on Instagram, LinkedIn, X, Facebook, and YouTube.
ABOUT ONIX
Onix is the world’s first Personal Intelligence® solution, a private, expert-driven AI built from the minds you’d most want in your corner to solve life’s hardest problems.
Onix doesn’t replace humans; it amplifies them. Participants get guidance from the experts they trust, anytime, anywhere, with personal, private advice. Onix gives experts a way to license their knowledge intentionally and maintain governance over how their intelligence systems operate, while scaling their work to help more people without surrendering their IP to centralized AI platforms.
PRESS CONTACT
Sara White
press@onix.life
832-314-2640
View original content to download multimedia:https://www.prnewswire.com/news-releases/onix-launches-tiny-labs-to-build-small-language-models-around-individual-expertise-302890965.html
SOURCE Onix

