×
Onix
Per-expert small language models, licensed from the expert rather than scraped
Last verified September 10, 2026 · Updated daily
What Onix is building
Onix trains a separate small language model per expert, on that one expert's own corpus, and calls the result an onix. The corpus is the part that matters: unpublished clinical protocols, working papers, case notes, proprietary frameworks, the judgment someone accumulated over a career and never put on the internet. Nicholas Nadeau's framing is direct — 'you can't scrape that; you have to earn it' — and earning it means a licensing deal, not a crawl. The architecture follows from the data deal. Models are small and run locally, chat data is stored on the user's device, and no conversation logs are kept, so there is no mechanism for data to reach a data centre in the first place. Bennahum claims inference costs around one two-hundredth of a frontier model. Nadeau's line is 'we don't want our users' data. We don't have a data bottleneck because our experts bring us all the data,' which is the rare privacy pitch that is also a unit-economics pitch. The business is a two-sided marketplace with an explicit split: experts set their own subscription price and take 70%, Onix takes 30%. More than 20 founding experts are live — William Li, Mark Sisson, Christian Drapeau, McCall McPherson, psychiatrist David Rabin, sleep psychologist Aric Prather, and others — across longevity, sleep and stress, food as medicine, pediatrics, and relationships. Users describe a problem, get matched, and talk to the relevant onix free before subscribing. Onixes can join group chats with each other and with human friends and family. Onix is explicit that it does not provide medical advice and has guardrails against drifting there. The company holds a trademark on Personal Intelligence, has a partnership with Mila, and was named to Canada's Top 100 AI Startups for 2026 by ALL IN.
Why this matters
Consumer trust in AI is the constraint this company is built around, and the Canadian numbers are stark: a 2025 poll found 83% of Canadians worried about AI and privacy, and a May poll found 46% had already asked a chatbot for medical advice while 40% believed AI would make Canadian healthcare worse. People are using these tools for health regardless of whether they trust them. Bennahum's read is blunt: 'AI is the most disliked new technology ever polled, probably since atomic weapons. And I think rightfully so.' The second argument is supply-side and probably the stronger one. Dave Rabin, one of the experts, put it as: 'All the work we put into our careers, the biggest problem is it never reaches the people who need to hear it.' Every practitioner with a career of unpublished material is watching frontier labs monetize an averaged version of their field while their actual work sits in a drawer. A 70% revenue share with retained IP governance is a concrete alternative, and if the model works, the expert supply compounds. The risk is honest and worth naming: expert credentials vary, wellness is a category with a long history of confident bad advice, and 'we don't give medical advice' is a guardrail that has to hold under real user pressure. That guardrail is an engineering problem, and it is a large part of what the technical roles here are actually for.
Open roles at Onix
6 positions we're tracking. Roles are re-checked daily and removed when filled.
Growth Marketing Lead (Old Montreal, in person)
First seen today
Growth Product Manager (Old Montreal)
First seen today
Lead Product Designer (Old Montreal)
First seen today
Brand & Communications Lead (Old Montreal)
First seen today
Expert in Residence (part-time, Old Montreal)
First seen today
Research Intern, Small Language Models — Mitacs (Old Montreal)
First seen today
Know when Onix is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Onix
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
Six roles posted on the public careers page, and BetaKit reports the 10-person company is hiring four more, mainly in GTM and sales, as it launches the paid model. The careers page is unusually explicit about the constraint: 'Small senior team. Old Montreal. In person.' Combined with the round's stated use for expert onboarding and research, this is a company that has told you exactly how many people it wants and where they have to sit.
Working at Onix
Onix is a Tech company based in Montreal, Canada. Working at a Tech company at this stage means opportunity to shape your role based on the company stage.
Montreal, Canada is where most Onix positions are located.
How to actually get hired at Onix
Why applying the normal way doesn't work
Onix runs an applicant tracking system, but hiring managers still work referrals first. A cold application to Onix isn't dead, it's just fourth in line behind internal referrals, sourced candidates and recruiter pipelines.
Who to contact at Onix
What to show them
The company just switched on a paid model for a product that is iOS-only and in early access, in a category where 83% of Canadians say they worry about AI and privacy. Build the acquisition plan that treats that distrust as the channel rather than the obstacle: which of the 20+ founding experts have audiences you can go through, what the creator-led UGC actually says, and what the first paid test is with a budget attached. Include the one message you would refuse to run, and why.
A cold email that works at Onix
What Onix screens for
Alpha Edison's Nick Grouf said plainly why they invested: 'The next generation of AI will be defined by trust. Onix has built a fundamentally different approach around human expertise and privacy.' Jeremy Zimmer co-founded UTA, which is the tell — this cap table reads the company as a talent-licensing business as much as an AI one, and the 70/30 split is the same shape as any representation deal. So the hiring window is a GTM window: the paid model just launched, the app is still iOS early access with Android to follow, and the four unposted hires are the ones who have to make a consumer subscription work. On the technical side the opening is narrower and more interesting. The Mitacs research role covers ingestion, agentic RAG, training, and the Expert Fidelity evals that grade whether a model is actually faithful to its expert. That eval problem is the company's real moat, and there is currently one posting on it.
Customize your CV for the Onix role. Matching the job description language helps clear ATS filters.
Don't make these mistakes
Do not pitch this as an AI wellness app, and do not tell Nicholas you are excited about small language models in the abstract. He was CTO of 1X and then of a data-labelling company, which means he watched frontier labs assemble their training sets from the inside; the privacy architecture here is a considered reaction to that, not a positioning exercise. David has been writing about the internet's bargains since he was a contributing editor at WIRED in the nineties and has already told a reporter that AI is the most disliked technology ever polled. Neither of them needs to be convinced that trust matters. What lands is the opposite: name a place where their own approach is hard. Expert fidelity evaluation, the failure mode where a small model confidently drifts into medical advice, or how you keep a per-expert model faithful when the corpus is a hundred hours of podcasts and forty pages of notes.
Mistakes that kill Onix applications
Generic CVs stand out at a -person company — and not in a good way. Fastest path to rejection.
Don't open with what you want. Open with what Onix is dealing with right now — Consumer trust in AI is the constraint this company is built around, and the Canadian numbers are stark: a 2025 poll found 83% of Canadians worried about AI and privacy, and a May poll found 46% had already asked a chatbot for medical advice while 40% believed AI would make Canadian healthcare worse, and what you'd do about it.
Most Tech applications get ghosted. A day-five follow-up can double your response rate.
Applying to Onix? Get the contact, not the form.
The Onix interview process
4 stages · 14 days typical · take-home: no · modelled from similar companies
We don't yet have verified candidate reports for Onix. What follows is the typical process for a -person Tech company — treat it as a model, not confirmed detail.
Interview stages
Recruiter Screen
Phone or video · 30 min
Basic qualification and logistics
Recruiter or HR
Hiring Manager Interview
Video call · 45 min
Role fit and experience deep-dive
Hiring manager
Technical/Functional Round
Video call · 60 min
Skills assessment and problem-solving
Team members
Final Round
In-person or video · 60 min
Culture fit and cross-functional alignment
Senior leadership
Onix interview timeline
Timeline: ~ days. That's about average than the Tech median (14 days at people).
Interviewed at Onix?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Onix interview experience →Onix jobs, frequently asked questions
How many jobs does Onix have open?
As of September 2026, Onix has 6 open positions.
Does Onix hire remotely?
Currently, Onix only has in-office roles in Montreal, Canada.
What roles is Onix hiring for?
Onix is hiring across Marketing, Product, Design, Other. The most recent opening is Growth Marketing Lead (Old Montreal, in person).
How do I apply for a job at Onix?
Use the apply links above, or check our guide to getting hired at Onix.
Does Onix respond to cold emails?
Not enough data yet on Onix's cold email response rates.
Who is the hiring manager at Onix?
At this size, hiring is usually run by the hiring manager.
How competitive is it to get hired at Onix?
Expect 100-250 applicants in the first two weeks for Tech roles at this size. Apply within 72 hours for best odds.
How many rounds is the Onix interview?
4 stages: Recruiter Screen, Hiring Manager Interview, Technical/Functional Round, Final Round.
Is the Onix interview hard?
Expect relevant experience and culture fit, not algorithm trivia. Candidates report Final Round as the toughest stage.
Does Onix give a take-home task?
No, Onix does not include a take-home stage.
How long does Onix take to get back to you?
Around 14 days across the full process.
What should I prepare for the Onix interview?
relevant experience and culture fit is the priority. Show you can work autonomously — that matters more than algorithms at people.
Where is Onix based?
Onix is headquartered in Montreal, Canada.
Get Onix roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Onix
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+