Yann LeCun joins Logical Intelligence as founding chair of research board for energy-based AI reasoning

US-based artificial intelligence (AI) startup Logical Intelligence has appointed Yann LeCun, former chief AI scientist at Meta, as the founding chair of its Technical Research Board, the company announced on January 22.

LeCun, one of the world’s most influential AI researchers and a Turing Award winner, left Meta late last year to launch his own startup, Advanced Machine Intelligence Labs, focused on building “world models” that can understand and navigate the physical environment. His decision to join Logical Intelligence signals a growing interest in alternatives to large language models for high-risk, real-world systems.

The company, founded by Eve Bodnia, also announced its flagship reasoning engine, Kona 1.0. A live public demonstration of Kona has been released on the company’s website.

Logical Intelligence is developing energy-based models (EBMs), a class of AI systems that reason by enforcing constraints rather than predicting probabilities. The company says Kona is designed to sit beneath modern AI stacks, acting as a verification layer for systems where failures are unacceptable.

“Kona is not a chatbot, assistant or generator,” the company said. “When software controls physical assets or financial risk, something else has to decide what actions are allowed before they happen.”

Unlike large language models that generate statistically likely answers, Kona evaluates all valid states of a system and ensures outputs remain within certified boundaries. Logical Intelligence says this makes the system suitable for applications requiring auditability, liability coverage and regulatory certification.

The company’s leadership team includes Fields Medalist Michael Freedman as Chief of Mathematics and ICPC World Champion and former Facebook engineer Vlad Isenbaev as Chief of AI. Patrick Hillmann, formerly a senior executive at General Electric and ex-Chief Strategy Officer at Binance, has joined as Chief Strategy Officer.

Logical Intelligence says Kona 1.0 can already solve complex reasoning problems such as Sudoku faster and with lower power consumption than leading large language models.

The company is positioning energy-based reasoning as a foundational layer for future autonomous systems, industrial automation and infrastructure AI.