Every paper links to its DOI on Preprints.org, and every article to LinkedIn. Within each topic the newest comes first. Paper dates are the dates each paper was submitted. The author list for each paper is on its Preprints.org page.
Declared world models for robots: a language model writes the rules of the scene, an engine runs them to plan and predict, and a monitor repairs the rules that prove wrong. This group also holds the audits of joint-embedding (JEPA) world models.
Studies of decoding backlogs in error correction, assembly of atom arrays, and protocol selection in continuously monitored quantum systems.
How the Great Pyramid could have been built: construction methods, ramps and logistics, studied with a discrete-event digital twin.
Articles on LinkedIn about running an enterprise on a live model of itself and its world, and on obligations instead of business processes.
AI agents work within obligations (a debtor, a creditor, a deliverable and a deadline) instead of fixed business processes, so the path can change while the commitment stays.
An enterprise builds an executable model of its operations and ecosystem, and makes decisions by simulation instead of through predefined processes.
Articles on LinkedIn about programming languages designed for coding agents and for AI applications, and about Poly.
Poly is a language for AI coding agents that combines formal verification, policy enforcement and durable execution, to catch the state-management defects that code review misses.
A survey of two families of languages created since 2023: some designed for coding agents to write, others for AI-integrated applications.