Progress
The goal of the development program at EpiCognix is to advance Ethical AI through principled, values-embedded design. This Progress page provides periodic updates on that work, highlighting key milestones from architecture design to embedding ethics into reasoning. Details are kept at high-level to safeguard intellectual property while ensuring adequate transparency for the purpose of providing progress updates.
Updates
Update – June 2025
The initial direction for the program was solidified. Core architecture was outlined, essential components were researched and identified, and foundational modules were integrated with a design that anticipates future requirements. The near-term target is a minimum viable framework, an early working system to validate the architecture before expanding its scope.
Update – August 2025
The ethical reasoning framework was developed and embedded into the AI’s reasoning pipeline. In parallel, a domain-agnostic integration architecture was established to support future applications across technical domains. These advances move the system closer to an early working version of the architecture, strengthening the foundation for an MVP demonstration before broader expansion. The focus is now on synchronizing files, consolidating modules, and identifying missing gaps needed for a complete build.
Update – September 2025
A provisional patent was filed for the Smart Model™ (short for Symbolic Memory and Reasoning Technology Model) architecture. This framework is designed to embed ethics into AI reasoning through symbolic memory and principled decision-making. The filing protects the architecture that’s been developing and marks an important step toward a defensible foundation as we continue building AI that is trustworthy, transparent, and aligned with human values.
Update – May 2026
During the work on system reasoning, it has become clear that AI memory storage strategy is critical. The primary function of the AI memory is to provide relevant history (or world experience) for a given context to the AI, so that its responses are applicable, appropriate, and sound. There are varying approaches to determine which world experience and details are important enough to retain for later recall. Identifying a solution that balances memory fidelity with hardware requirement will be the key to resolving this issue.
Update: A method for managing memory for efficient recall and traceability has been identified. it uses the memory fields already implemented in Smart Model.
Update – October 2026
EpiCognix is introducing a new development approach to support the continued advancement of Smart Model™ cognitive architecture.
While substantial progress has been made at the code level, these modules have not yet been fully integrated and evaluated as a unified system. Integrating the entire architecture simultaneously would introduce considerable complexity, making it difficult to isolate technical issues and systematically validate individual capabilities.
To address this challenge, EpiCognix is adopting a revised development methodology emphasizing incremental integration, testing, and refinement. Under this approach, individual cognitive capabilities will be progressively assembled into a functioning system, with each stage building upon previously tested components. A controlled, interactive virtual environment will provide a practical setting for evaluating the system’s behavior under changing conditions and accumulated experiences. This methodology is intended to reduce integration complexity, improve the identification and resolution of technical issues, and allow architectural refinements to be informed by experimental results throughout development.
The underlying system architecture and its long-term research objectives remain unchanged. The revised approach represents a practical refinement of the development process, intended to establish a more reliable path toward a fully integrated Smart Model MVP.
Future updates will document progress in system integration, functional testing, and the progressive validation of cognitive capabilities.