This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts

Onix, a new startup co-founded by former WIRED contributor David Bennahum, is launching a platform it calls a 'Substack for chatbots.' The service allows users to pay subscription fees to converse with AI clones, or 'Onixes,' of real human experts. These bots are trained on the expert's personal content and designed to mimic their knowledge and personality, delivering advice as if in a one-on-one consultation. The company is starting with a curated group of 17 vetted experts, with a current focus on health and wellness influencers and professionals.

Onix's core pitch is resolving common AI concerns: privacy, accuracy, and creator compensation. The company uses what it calls 'Personal Intelligence' technology, which stores encrypted user data locally on the device. Bennahum claims this means even if the Canada-based company receives a government request, it can only provide a user's email. The models are trained exclusively on the expert's own materials to avoid IP issues, and guardrails are implemented to limit conversations to the expert's domain, aiming to minimize hallucinations. However, early testing by WIRED showed the bots could still be led off-topic and produce inaccurate information.

The business model positions an expert's knowledge as a 'capital asset' that generates revenue independently of their time. For experts, it's a potential passive income stream; for users, it's purportedly more private and sourced advice than a standard LLM. The platform explicitly states these AI consultations are not replacements for actual medical treatment or therapy. While the concept of expert chatbots isn't new—psychologist Becky Kennedy's similar 'Gigi' bot helped her company earn $34 million last year—Onix is betting its privacy-focused, expert-centric platform will carve out a significant niche.