Ontological Language Models (OLMs)

OLM is a new language model alternative or complementary to LLMs.

Its main advantages are (1) the delivery of outputs without any hallucination (deterministic-AI), (2) reduction of the cost of LLMs and (3) enabling total content control (RAG) not allowing manipulated language driven by statistical bias. Additionally, OLM offers instant troubleshooting/updates due to its transparency.

OLM+LLM hybrid implementation - OLMs replacing the output functions while LLMs operate at the input layer - manifest all the advantages in high-risk industrial applications. This is particularly true where enterprises demand to use their own content via RAG configuration.

OLM is based on an event-based-ontology where event-concepts take the center role. A new machine learning is invented to automate ontology development and merging to backgorund ontology for a given content. No coding, prompting, or expertise is required. All such capabilities are observable in OLM Sand Box.

OLM Version 1.1

Version 1.1, released September 5, 2026, works as a general background ontology plus an ingestive, dynamic ontology for a given content. It can be fully tested in our Sand box. Please request access to the Sand Box using the form below.