Like many of you, I’m deeply immersed in learning and using AI. It’s not an exaggeration to say it has been life-changing for me.

Right now, AI is largely synonymous with Large Language Models — ChatGPT, Gemini, Perplexity, Claude (my current favorite), and even Grok. But I suspect that in the not-so-distant future, we’ll look back on these models the way we now look back on the first search engines and portals: Yahoo, Lycos, AltaVista, and AOL. Useful in their day, but just the opening chapter.

The LLM companies are on an extraordinary roll. Yet it’s easy to forget that most of the data on the internet isn’t actually available to them likely less than 10%. LLMs are aggregators of knowledge scraped from publicly available information, supplemented by what they learn from our interactions. Powerful, but inherently limited.

I’ve been thinking for some time about how AI will evolve, and especially how it will become more personalized. So I was excited when my good friend Nick Grouf — one of the founders and managing directors of the Alpha Edison Fund, where I’m a limited partner told me about Onix.

Onix is taking a very different approach to AI — one that focuses less on scale and more on trust. Their newly launched platform, Personal Intelligence®, flips the typical AI model on its head. Instead of relying on large, generalized systems trained on the entire internet, Onix builds individualized AI systems trained on the proprietary knowledge of specific experts. The idea is simple but powerful: rather than an “average answer,” users get guidance rooted in the thinking of an expert they actually trust.

The model is especially compelling in health and wellness, where Onix is launching first. Users subscribe to an expert’s “intelligence system,” built from that expert’s own research, frameworks, and intellectual property — often including material that isn’t publicly available. Over time, the system learns from ongoing interactions, creating a longitudinal relationship in which the advice improves as it better understands you. Importantly, this is positioned as a complement to professional care, not a replacement, with safeguards to escalate serious issues when needed.

What really differentiates Onix is its architecture around privacy. Conversations are end-to-end encrypted and stored locally on the user’s device, not on centralized servers. Personal data lives in a private vault that even Onix itself cannot access. In a world where AI systems increasingly depend on harvesting and monetizing user data, Onix is betting that trust — not compute — is the real bottleneck. Given declining public confidence in AI for sensitive domains like healthcare, that may be exactly the right insight.

At launch, the platform includes a notable group of experts such as William Li and Jeffrey Bland, among others across medicine, nutrition, and mental health. Their participation signals another important shift: a new model in which experts can scale their knowledge without giving up control of their intellectual property.

I was one of the first people to be “cloned.” It was a strange experience interacting with my digital twin. I’d come up with a question, answer it myself, and then ask my twin the same question and compare notes. The answers were remarkably close.

I think this is a big deal. There are so many experts whose thinking I’d love to access, but they can’t possibly scale themselves — so instead they give talks, publish articles, and write books. Valuable, but nothing like the experience of actually being in a conversation with them.

With Onix, you can now access some of the most talented people in the world for less than the cost of booking a 30-minute session of their time. If it succeeds, it could mark the beginning of a more personalized and more private era of AI.

Check out their app.

---