Company
A small team. A long horizon. An unambiguous position.
NeuroBazar is an AI infrastructure team in Ottawa, Canada. We design and ship the Midcore platform — twenty modules, twelve products, four patent families — and we hold a position on the era that we are willing to be measured against.
Mission
Make the consenting AI the easier AI to ship.
We were founded on the conviction that the AI era is not a software era — it is a social one. The defaults of the consumer surfaces being built now will shape the next twenty years of how people relate to institutions, to information, and to one another. Those defaults are extractive by gravity: the cheapest shape to ship is the one that treats the user as a data source. The mission is to build the infrastructure that makes the other shape — consenting, cited, accountable, sovereign — cheaper than the extractive one.
The work is intentionally larger than the team. We file patents because the next decade’s infrastructure deserves the same legal scaffolding the previous decade’s did. We publish manifestos and essays because trust without checkability is not trust. We ship products because the market is the mechanism that will decide which substrate becomes the default. We hold a public Charter because we expect to be held to it.
2023
Founded in Ottawa, Canada.
4
Patent families filed across US, CA, USPTO.
9
Midcore provisionals drafted.
20
Charter modules engineered for the dual purpose.
Why small
Small by design, not by stage.
The temptation in this market is to be a hundred-person firm by year three. We have rejected that path on purpose. A small senior team can ship a hundred-person product if the work is concentrated; it cannot ship a thousand-person product, and we do not want to. The product we are shipping is infrastructure — twenty modules, made well, documented in public. The mass of work is in the substance, not in the headcount.
Staying small is also how we keep the position honest. The larger the firm, the harder it is to refuse the extractive shortcut on a Friday afternoon when the quarter is closing. Small lets us refuse. Small is, in this domain, an ethical posture as much as an operational one.
Principles
Six operating principles. The same ones we publish.
Citation by default
Every AI-generated claim cites the source it came from. If we can’t cite it, we don’t ship it as fact.
Consent before action
No agent acts on money, messages, or data without explicit, scoped permission in the live session.
Refusal as a feature
The user can turn personalisation, recommendations, and classification off — and the product still works.
Audit by default
Every meaningful AI action leaves a hash-chained record. Customers own it. Users can request it.
Sovereignty by default
Data, evidence chains, and fine-tunes are exportable. No economic moat made of switching cost.
No dark patterns
We refuse to ship anything whose value depends on the user misunderstanding what is happening.
Talk to us