Binary Kinetics filed 1 patent application in H1 2026, covering AI & Machine Learning (1).
The filing details a stateless AI Game Master architecture designed for interactive fiction systems. This approach uses transaction-based context reconstruction to reduce per-session memory requirements by 99% while incorporating multi-provider LLM failover capabilities for scalability.
In AI and ML, Binary Kinetics has 1 filed application centered on a new architectural approach to running AI game masters at scale. Rather than storing full conversation histories between sessions, the system reconstructs the context an LLM needs by replaying compact action logs, drawing on principles from event sourcing but adapting them specifically to the constraints of LLM context windows. This is a meaningful departure from how event sourcing typically works in databases, since the goal here is not to rebuild application state but to regenerate a semantically coherent prompt. The architecture also abstracts away dependence on any single LLM provider, allowing the system to shift between providers without disrupting the game mechanics layer that enforces consistent rules across sessions.
All data sourced from USPTO patent filings. Google Patents may take several weeks to index recent publications. If a link is unavailable, search for the patent number at USPTO Patent Public Search.