Our models

Isaacus offers some of the most accurate, efficient, and scalable legal models in the world. This page provides an overview of those models, including their capabilities and constraints. If you’re looking to understand the pricing of these models, please refer to our pricing page. If you want to use these models in practice, check out our quickstart guide. All Isaacus models are accessible through our API as well as via air-gapped deployments on Amazon SageMaker.

​

Enrichment

Enrichment is the process of transforming unstructured documents into rich, structured, hierarchical knowledge graphs representing key sections, concepts, entities, and their relationships to one another. Enrichment is a critical first step in preparing data for use in downstream applications such as agentic search, eDiscovery, and retrieval-augmented generation (RAG) systems demanding high-quality structured inputs. Our flagship enrichment model is Kanon 2 Enricher.

​

Embedding

Embedding or vectorization is the process of converting content into sets of numbers that, when compared mathematically with each other, quantify how similar they are in meaning. Embeddings are used to power semantic search engines, text classification, and cluster analysis as well as the retrieval component of retrieval-augmented generation (RAG) applications. Isaacus’ flagship embedding model is Kanon 2 Embedder, which currently ranks as the most accurate legal embedding model on the Massive Legal Embedding Benchmark (MLEB).

​

Reranking

Reranking refers to the process of scoring and sorting a set of documents based on their relevance to a particular query. Reranking is typically used in conjunction with a search engine, where the search engine retrieves a set of documents based on a user’s query, and the reranker scores and sorts those documents to present the most relevant ones to the user (or, when used as part of a retrieval-augmented generation (RAG) pipeline, to present the most relevant documents to a generative model). Our flagship reranker is Kanon 2 Reranker, which ranks as the most accurate reranker on Legal RAG Bench. All of our universal classifiers can also be used as rerankers.

​

Extractive question answering

Extractive question answering or answer extraction is the process of extracting answers to questions from documents. Extractive QA can be used to solve a broad range of information extraction problems from pulling out key dates from a contract to highlighting legal doctrines relied upon in a judgment. We offer the Kanon Answer Extractor extractive QA model, optimized to balance precision with throughput.

​

Universal classification

Universal classification (also known as zero-shot classification) refers to the process of determining whether a given statement about a document, for example, “this is a confidentiality clause”, is supported by that document. Unlike traditional classifiers, universal classifiers don’t require training data to classify a document. In practice, universal classifiers can be used for a broad range of information retrieval and extraction tasks, from pulling out particular types of clauses from contracts all the way to scoring the relevance of documents to search queries. Our flagship universal classification model is Kanon Universal Classifier.