Transforming Legal With Generative, Extractive, and Agentic AI

In a recent LawVu article, Marija Barbarich reports on how in-house legal teams are transforming the legal field with generative, extractive, and agentic AI. The article explains how the three types of AI offer innovative ways to streamline processes and improve decision-making. 

Generative AI, built on large language models, generates human-like responses. Because it predicts “likely” responses, it has the potential to produce hallucinations or inaccurate content, often requiring validation. Extractive AI, by contrast, identifies and retrieves precise data from existing datasets, ensuring accuracy and traceability while mitigating concerns about copyright and plagiarism. These characteristics are crucial in compliance-intensive tasks such as contract review or invoice analysis. 

The two types of AI serve different yet complementary purposes in legal workflows, with generative AI suitable for creative drafting and answering questions, and extractive AI excelling in data extraction and summarization. 

The conversation doesn’t end there, however. Agentic AI is emerging as a new frontier operating with self-prompting capabilities. Unlike its generative and extractive counterparts, agentic AI can handle multi-step tasks such as automated contract generation, risk analysis, and due diligence with minimal human intervention. This positions it as a transformative force for legal teams seeking greater efficiency and strategic leverage.

As in-house legal teams are transforming the legal field with generative, extractive, and agentic AI, understanding their differences is crucial. Each offers distinct benefits and risks, particularly in areas like compliance, reliability, and user interaction. By aligning AI capabilities with specific legal tasks, in-house teams can unlock significant productivity gains while managing risk and upholding ethical standards.