Why we built our own model
Most “AI companies” are a thin wrapper over someone else’s API. That’s fine until your data has to leave the country, your costs scale with every token, or the model simply doesn’t understand how a business here actually works. So we built our own.
Lux 1.0 is our first foundation model — designed, trained, and tuned in-house. It’s the engine underneath the agents and systems we ship, and it’s built for one thing: doing real work for real businesses, reliably, in production.
Sovereign by default
Lux runs where you need it — our managed cloud, your own VPC, or fully on-prem. Your documents, your customers, your numbers: none of it is sent to a third-party model provider. Data residency stays in New Zealand or Australia, full stop.
For regulated industries — finance, healthcare, government — that isn’t a nice-to-have. It’s the difference between “we’d love to” and “we can actually ship this.”
Tuned to your domain
A general model is a generalist. Lux is built to be fine-tuned on your domain — your documents, your workflows, your tone of voice — so it answers like someone who actually works at your company, not a chatbot guessing.
We pair it with retrieval and tool-use so it can ground every answer in your own sources and take real actions, with a human in the loop on anything that matters.
What’s next
Lux 1.0 is in production with our first partners now. We’re already training the next version — bigger context, faster inference, and tighter evals so you can see exactly how it performs on your tasks before it goes live.
If you want a private, tuned model that keeps your data yours, this is the one.