A FRAMEWORK FOR THE ADOPTION OF LEGAL LLM AI IN THE INDONESIAN CRIMINAL JUSTICE SYSTEM (CORRUPTION) BASED ON TOGAF: TRANSFORMING LEGAL WORKFLOW EFFICIENTLY

Abstract

This study creates an architecture for incorporating Large Language Model (LLM) Artificial Intelligence into Indonesia's criminal justice system for corruption cases, employing The Open Group Architecture Framework (TOGAF) to improve legal workflow efficiency. The study applies qualitative research methodologies, analyzing over 35,000 Supreme Court corruption case verdicts, prescriptive analysis and conducting Focus Group Discussions (FGDs) with legal and IT specialists. The framework is validated using proof-of-concept testing on Claude 3.5 Sonnet. The findings show that a TOGAF-based framework successfully addresses technological difficulties and ethical issues using a layered infrastructure approach, whereas the Retrieval Augmented Generation (RAG) architecture processes legal documents and improves workflow efficiency. The research developed a solution architecture for integrating LLM AI with current court systems via SPLP (Sistem Penghubung Layanan Pemerintah) interface points. The study concludes that the recommendation for a structure may considerably enhance Indonesia's legal workflow efficiency in corruption cases by integrating AI in a systematic integration keeping security and ethical requirements. The study suggests creating thorough testing frameworks, applying in place rigorous validation processes, and creating mentorship programs to encourage legal professionals to utilize technology.