I sit on both sides of this
I’m the CEO of Ulter, and I’m also a user of AI products myself, so I understand the pull from both directions. Learning from users’ chats is a legitimate, real method for making a model better — I’m not going to pretend otherwise.
Learning is not advertising
But learning in order to improve a model is a different thing entirely from using that same data for advertising. Those are two different purposes, and they deserve two different products.
Learning in order to improve a model is a different thing entirely from using that same data for advertising.
Who ElaNa is for
That’s why Ulter will offer both the Okran LLM and ElaNa rather than forcing everyone into one model. ElaNa is built specifically for the people and organizations for whom privacy isn’t optional: individual users who simply want that peace of mind, corporations working on projects they can’t afford to have leaked, and even governments. For all of them, running fully offline — with no dependency on a live connection to a server that could see their data — is the actual requirement, not a nice-to-have.
The full ElaNa research entry →
How Okran.ai, the Okran LLM and ElaNa differ →