The neural network architecture underlying modern LLMs (GPT, Claude, etc.) that enables contextual understanding across long documents; has dominated legal AI since approximately 2020.
Last reviewed: 2026/05/19
Anchoring AI-generated text in specific retrieved source documents, reducing hallucination; a grounded response cites the specific passage supporting its claim.
Tech / ModelA subset of machine learning using multi-layered neural networks that powers contract clause extraction, semantic search, and LLMs; modern legal AI tools are predominantly deep learning systems.
Tech / ModelAn LLM deployed exclusively for one organization with no data sharing with other customers or the model provider for training; provides stronger confidentiality guarantees at higher infrastructure cost.
Thomson Reuters' GPT-backed legal research and drafting with Westlaw integration (relaunched as CoCounsel Legal, 2025).
Enterprise AI for portfolio-level contract analysis and institutional memory.
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.
Legal department workflows: contract lifecycle, regulatory tracking, outside counsel management, and risk.
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.
A transformer model is the neural network architecture, introduced by Google researchers in the 2017 paper "Attention Is All You Need," that underlies the current generation of large language models (LLMs) — including GPT, Claude, Gemini, and the models powering most modern legal AI tools. The transformer's key innovation is the attention mechanism, which allows the model to consider the relationship between all tokens in a sequence simultaneously — enabling contextual understanding across long documents rather than processing text word-by-word. Legal AI tools built on transformer architectures understand clause relationships within contracts, recognize that a defined term in section 1 affects meaning throughout the document, and generate coherent legal text.
The transformer architecture is the reason that modern legal AI tools are qualitatively more capable than their predecessors. Earlier document analysis tools used pattern matching and simpler statistical models that could not understand context; transformer-based tools understand meaning in context, enabling capabilities like semantic search, clause relationship analysis, and coherent drafting that keyword-based tools cannot approach.
Lawyers evaluating AI tools should understand that most serious legal AI tools are now built on transformer-based LLMs — either general-purpose models (GPT, Claude) fine-tuned on legal data, or purpose-built legal models. The quality of the underlying model and the quality of legal fine-tuning both affect performance. A powerful base model poorly fine-tuned for legal tasks can underperform a less powerful model with excellent legal fine-tuning.
The context window — how many tokens the model can process at once — is a practically important transformer characteristic for lawyers. A model with a small context window cannot process a 100-page agreement in a single pass; a model with a large context window can. Context window size affects what tasks a model can handle on long legal documents.
Harvey is built on frontier LLMs fine-tuned on legal data, using the transformer architecture's long-context capabilities to process lengthy legal documents in unified context rather than splitting them. CoCounsel similarly applies transformer-based LLMs with legal domain tuning to its research, drafting, and review task suite.
Luminance applies transformer-based models specifically for contract analysis, with training focused on commercial contract language and clause relationship understanding across long agreements.