Noeon Research is a startup developing an AI architecture as an alternative to LLMs for solving a variety of problems. The architecture could be a core component for products aimed at automated development and completion of IT projects, producing industrial designs, or creating neural networks on demand.
Currently, Noeon Research is concluding its work on the Proof-of-Concept. In our research, we are combining Machine Learning approaches that allow the efficient processing of unstructured data and Symbolic approaches that provide a strong inductive bias via efficient knowledge encoding.
The architecture's robust multi-step reasoning offers reliable solutions for complex scientific and industrial applications. The architecture's interpretability allows the operator to examine the process of solution generation. The system is also capable of interacting with the environment or the operator to request additional information to deal with uncertainty or lack of knowledge.