(Portfolio Profile) Onix

Key takeaways

Onyx enables experts to create personalized AI models based on their private knowledge, which can then be shared with millions in a trustworthy manner.
The platform emphasizes privacy, encryption, on-device processing, and consent-based data use to build models that truly represent expert judgment and personality.
A key competitive advantage is the focus on domain-specific models that are more accurate, reliable, and aligned with expert values than general-purpose large AI models.
The revenue-sharing model incentivizes experts to participate and maintain quality, with ownership of the models retained by the experts.
The future of AI in this space involves highly specialized, trustworthy avatars that serve high-stakes domains like health and wellness, rather than broad, unfocused models.
Building trust and aligning interests with experts is fundamental to creating effective and ethical AI tools.
Listeners interested in joining the platform or participating as experts can sign up at onex.life, and use the access code 'ride' for entry.

Summary

Introduction to Onyx and the Founders' Vision

The episode introduces Onyx, a startup that allows experts to train their own AI models, tailored to their domain expertise, which are then accessible to millions. Founders Nicholas Nadeau and David Bennahum discuss their aim to embed trust and specificity into AI, emphasizing that their models are built on a fundamentally different philosophy from general-purpose models. They highlight the importance of creating domain-specific AI avatars that represent true expertise, complete with personality and judgment, built with permission and based on experts’ private data, manuscripts, notes, and implicit knowledge.

The Technical and Philosophical Foundations of Onyx

The founders explain their focus on domain-specific small language models as a way to ensure fidelity, accuracy, groundedness, and a representation of personal judgment. Unlike large, general AI models trained on vast, unfocused datasets, Onyx's models are based on content curated with explicit expert permission. A core belief is that AI should serve human flourishing, not surveillance or extraction, with an emphasis on privacy, encryption, and on-device data processing to ensure user control and trust.

Business Model and Expert Participation

Onyx’s revenue model shares 70% of earnings with the experts whose knowledge is embodied in the models, fostering alignment of interests. The platform is selective, actively recruiting top-tier health and wellness domain experts, and plans to expand into adjacent categories like self-improvement. Experts retain ownership of their models and can leave the platform anytime, ensuring integrity and control. The approach emphasizes partnership over vendor-client relationships, with a focus on scalability and high-quality expertise.

Use Cases and Future Applications

The platform supports a broad range of practical applications, such as personalized health advice, mental health support, family care, and multi-party group chats involving experts and users. The hosts discuss how these targeted AI-driven interactions could provide more accurate, trustworthy, and personalized experiences, especially in high-stakes areas like healthcare. They envision a future where AI models are more transparent, domain-specific, and on-device, reducing costs and increasing privacy.

The Broader Context and AI Competition

The founders draw parallels to the early days of the internet, noting that this AI revolution shares similarities with the 1990s, marked by both high optimism and anxiety. They emphasize trust as a key differentiator, criticizing large general models for lacking expert-backed fidelity. Their approach advocates for highly specialized models to challenge the notion that a single AI can be universally competent, aligning with a humanist vision that values decentralization, trust, and expert integrity.

Personal Backgrounds and Motivations

Nicholas Nadeau and David Bennahum share their journeys: Nadeau's extensive background in robotics, biomedical engineering, and AI, and Bennahum’s roots in media, digital strategy, and a long history of working at Wired and in digital innovation. Both express a shared humanist motivation—using technology to improve lives and uphold trust and privacy—driving their investment in Onyx. Their stories highlight a commitment to building responsible AI that respects individual and expert rights.

Closing Remarks and How to Get Involved

The episode concludes with practical information for listeners interested in the platform or becoming experts. Onyx is building in health and wellness but plans to expand into self-help and related fields. Interested experts can sign up via onex.life, and listeners can access a special trial using the code 'ride'. The founders invite collaboration and emphasize the importance of community in this pursuit of trust-driven, domain-specific AI.

Show notes

In this episode, we explore Onix, a platform that enables experts to train their own AI models to help millions with health, wellness, and personal development. Founders Nicholas Nadeau and David Bennahum discuss how their domain-specific, trust-based approach is reshaping AI and expertise sharing.