Title: An Autonomous Agentic-RAG Framework for Statutory Analysis within the Tanzanian Judiciary
Authors: Kenneth Brown, Michael Kinyunyu, Joseph chacha, Yacob Norvat, Yassir Salum, Witness Mgana, Witness Suleiman, Sir Yusuph koni
Volume: 10
Issue: 4
Pages: 46-50
Publication Date: 2026/04/28
Abstract:
The Tanzanian judiciary faces persistent challenges including case backlogs, slow judgement delivery, and inconsistency in the application of legal precedents - problems shared across Sub-Saharan Africa. Traditional judicial drafting relies heavily on manual research of laws, statutes, and prior rulings, a process that is time-consuming, error-prone, and dependent on the availability of experienced legal professionals. This paper proposes and reviews the design and development of an AI-Powered Judgement Drafting System that leverages Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) to assist judges in producing structured, legally accurate judgement drafts. The system retrieves relevant legal precedents and statutes from a vector-indexed knowledge base of Tanzanian and East African court records, then generates contextually grounded drafts through a fine-tuned LLM. A web-based interface enables judges and legal practitioners to interact with the system. This paper examines existing AI and NLP approaches in the legal domain, evaluates their architectural choices, limitations, and alignment with the Tanzanian legal context, and situates the proposed system within this body of literature. Key gaps identified include the lack of Africa-specific legal AI datasets, limited deployment of RAG pipelines for low-resource judicial settings, and insufficient attention to citation verification and legal traceability. The proposed system addresses these gaps through region-specific data collection, a RAG pipeline tuned for Tanzanian law, a citation verification module, and iterative usability testing with legal practitioners.