July 21, 2026
Open-weight models now catch the frontier in weeks: Kimi K3 matched Fable 5 in five weeks. Icosa gives regulated firms frontier-level performance from far smaller domain-expert open models via fine-tuning and Combinatorial Reasoning.
July 20, 2026
Usage limits, not the work, decide when you stop. Why agentic AI made metered intelligence far more expensive, how the limits actually work, and what owning your model changes.
June 29, 2026
The shutdown of Fable, and more recently, the delay of OpenAI's GPT-5.6, show that frontier AI is coming under government control and can disappear without warning. For most enterprise work, a private, custom model you own performs just as well, while saving money and ensuring data privacy.
June 19, 2026
Combinatorial Reasoning automates chain-of-thought by generating many candidate reasons and using combinatorial optimization to select the best subset. A look at how the method works, the physics-inspired solvers behind it, and how Icosa integrates it into fine-tuning local, domain-specific models for frontier-level reasoning at lower cost and greater privacy.
May 14, 2026
AI ownership is one of the most consequential decisions a business can make right now. A breakdown of the three main deployment models (public cloud AI, open-source AI, and on-premises private AI) and what it really means to own your AI.
January 12, 2026
Why local, domain-trained models outperform general-purpose systems in accuracy, privacy, and control. How domain-specific Small Language Models trained for particular fields can match or exceed the performance of much larger general-purpose systems while providing greater reliability, efficiency, and control.
November 3, 2025
Exploring how domain-specific Small Language Models running on-device achieve comparable performance while using 99% less energy and eliminating water consumption compared to cloud AI.
Exploring the economic advantages of private AI deployment over public cloud services, including cost savings, data sovereignty, and competitive advantage.