LAW: Legal Agentic Workflows for Custody and Fund Services Contracts (2412.11063v1)

Abstract: Legal contracts in the custody and fund services domain govern critical aspects such as key provider responsibilities, fee schedules, and indemnification rights. However, it is challenging for an off-the-shelf LLM to ingest these contracts due to the lengthy unstructured streams of text, limited LLM context windows, and complex legal jargon. To address these challenges, we introduce LAW (Legal Agentic Workflows for Custody and Fund Services Contracts). LAW features a modular design that responds to user queries by orchestrating a suite of domain-specific tools and text agents. Our experiments demonstrate that LAW, by integrating multiple specialized agents and tools, significantly outperforms the baseline. LAW excels particularly in complex tasks such as calculating a contract's termination date, surpassing the baseline by 92.9% points. Furthermore, LAW offers a cost-effective alternative to traditional fine-tuned legal LLMs by leveraging reusable, domain-specific tools.

• The paper presents the LAW framework that leverages domain-specific tools and text agents for efficient legal contract analysis.
• It demonstrates a 92.9 percentage point improvement in tasks like calculating contract termination dates and supports complex multi-hop reasoning.
• The framework spans 23 years of regulatory contracts, offering a scalable and cost-effective solution for legal document management.

An Overview of LAW: Legal Agentic Workflows for Custody and Fund Services Contracts

The paper, "LAW: Legal Agentic Workflows for Custody and Fund Services Contracts," introduces an innovative framework, LAW, designed to enhance the interaction with legal contracts specific to custody and fund services. These contracts are essential as they delineate key provider responsibilities, fee schedules, and indemnification rights, among other elements. The authors identify the challenge of utilizing off-the-shelf LLMs for such tasks due to the inherent complexity and unstructured nature of these documents, which often contain dense legal jargon beyond the context windows of existing LLMs. To address these challenges, LAW offers a modular design, employing a suite of domain-specific tools and text agents to process user queries efficiently.

Key Contributions and Results

The paper highlights three primary contributions of the LAW framework:

1. Modular Approach Utilizing Domain-Specific Tools: LAW leverages legal domain-specific tools and text agents to tackle retrieval and analytical tasks within legal documents. This approach also accommodates user queries ranging from direct data retrieval to complex multi-hop reasoning.
2. Performance Metrics: The empirical results demonstrate that LAW substantially outperforms the baseline models in complex tasks like calculating a contract's termination date, achieving a significant improvement of 92.9 percentage points over the baseline. The system showcases its efficiency in answering retrieval-type queries with high accuracy and analytical-type questions through enhanced contextual understanding.
3. Comprehensive Regulatory Scope: The framework covers twenty-three years of regulatory contracts, embodying a robust solution for performing retrieval and analytical tasks that require understanding across multiple documents.

Technical Implementation

The LAW framework adopts a sophisticated orchestration methodology by harnessing a range of domain-specific tools designed to manage tasks like extracting relevant contract information or computing the life cycle of contracts. These tools are adaptable and constructed to be reusable across different contracts, underscoring their cost-effectiveness as an alternative to traditional fine-tuning approaches.

Furthermore, text agents are employed to deliver insightful analytics such as summaries and comparisons of legal clauses, handling sections which may span the LLM's context limit efficiently through a partitioned processing approach.

Implications and Future Directions

The innovation brought forward by LAW holds significant practical implications by enhancing the automated processing of legal contracts. It presents a paradigm shift from reliance on purely fine-tuned LLMs to a more interactive, modular system capable of leveraging both open-source and closed-source LLM capabilities. This advancement has the potential to streamline contract management processes in the financial sector, optimizing efficiency and accuracy in interpreting complex legal documents.

The theoretical implications suggest a broader application of orchestrated agentic systems beyond the financial sector, catering to various document-heavy industries where traditional LLMs fall short. Future research could explore LAW's application in non-English legal texts, expanding its global usability, and integrating more specialized agents tailored to other specific domains.

In conclusion, the introduction of LAW as a framework for processing custody and fund service contracts exemplifies a significant advancement in Legal AI, providing a robust structure for domain-specific contract analysis and management. By marrying domain-specific tools with powerful LLMs, LAW presents a comprehensive solution for overcoming critical challenges in legal document interpretation and management.