An embedding is a numerical vector representation of text — such as a word, sentence, or document — produced by a machine learning model, enabling AI systems to measure semantic similarity between texts and retrieve relevant information.
Last reviewed: 2026/05/19
A large language model (LLM) is an AI system trained on large volumes of text data to predict and generate human-like text; it serves as the core engine underlying most legal AI tools for research, drafting, and document analysis.
Tech / ModelRetrieval-Augmented Generation (RAG) is an AI architecture that combines a retrieval system — which fetches relevant documents from a specified corpus — with a generative language model that produces answers grounded in those retrieved documents, rather than relying solely on the model's training data.
Tech / ModelVector search is a retrieval method that finds documents semantically similar to a query by comparing numerical vector representations (embeddings) rather than exact keyword matches, enabling natural language queries to surface conceptually relevant results.
AI-powered legal research with citation-validated answers from Westlaw.
Conversational legal research with real-time Shepard's citation validation.
Cloud eDiscovery with AI predictive coding and document summarization.
AI legal research pioneer (CARA AI); standalone retired 2025, its technology now powers Thomson Reuters CoCounsel.
Move from this definition to role-based legal AI shortlists and the selection criteria that matter for each type of legal team.
Am Law 200 and global firm workflows: accuracy at scale, security compliance, and matter-level auditability.
Last reviewed: 2026/05/19. Definitions are written by the LawyerAI Editorial team. Commercial relationships are disclosed and do not determine editorial scores or conclusions. See our Sponsorship & Affiliate Disclosure.
An embedding is a numerical vector representation of text — such as a word, sentence, or document — produced by a machine learning model, enabling AI systems to measure semantic similarity between texts and retrieve relevant information.
Embeddings are the technical mechanism that enables modern legal AI tools to find semantically relevant documents even when the exact keywords don't match. Traditional Boolean search requires the query terms to appear in the document. Embedding-based search understands that "breach of warranty" and "warranty non-conformance" describe similar concepts, and that a case discussing "reasonable reliance" may be relevant to a query about "detrimental reliance" — even if the exact phrase differs.
For lawyers using legal research AI, this means the tool can surface relevant cases that would have been missed by a keyword search. A contract lawyer searching for cases discussing a novel contractual term can find conceptually related precedents that use different terminology. This expands research coverage and reduces the risk of missing controlling authority.
In e-discovery, embeddings enable conceptual clustering — grouping documents by their semantic content rather than just keyword overlap. This allows reviewers to identify a responsive document population based on conceptual relevance, supplementing traditional keyword search protocols.
Understanding embeddings helps lawyers evaluate what a tool's "semantic search" claim actually means and why two tools may return very different results from the same natural language query — the underlying embedding model significantly affects retrieval quality.
Embeddings are used throughout legal AI tools, primarily in the retrieval component of RAG systems. When a lawyer submits a query to a legal research AI, the query is converted into an embedding vector; the system then finds the documents in its corpus whose embeddings are most similar, and those documents are passed to the LLM to generate an answer.
The quality of the embedding model determines the quality of the retrieval step. Tools that use embedding models specifically trained on legal text tend to produce better semantic matches for legal queries than those using general-purpose embedding models.
Westlaw Precision AI and Lexis+ AI have invested heavily in legal-domain embedding quality to ensure their retrieval correctly identifies legally relevant materials across their large content databases. E-discovery platforms like Everlaw and Relativity AI use embeddings for conceptual search and near-duplicate detection.
Embedding quality is not directly visible to end users but can be assessed by testing edge cases: whether the tool finds conceptually related materials that share no keyword overlap with the query.