Initiative overview
The JULIO AI Platform is Peru's most ambitious digital transformation project in the public sector, designed to enhance registry services using artificial intelligence. Rather than replacing human judgment, it serves as a tool to optimise legal processes through three main pillars:
- Pillar 1: Specialised Consultation Modules: Use natural language processing (NLP) and large language models (LLMs) to provide structured access to decades of legal precedents, technical resolutions, and plenary agreements.
- Pillar 2: Learning and Study Centre: Provides practical resources for staff, including AI manuals, optimised prompt libraries, and adaptive templates. It features an automated generator for official observation and rejection notices (esquelas) to ensure compliance with standards.
- Pillar 3: Integrated AI Access: Offers direct access to major AI applications with specialised guides tailored to specific registry contexts.
Results, outcomes and impacts
The main outcomes of the JULIO AI Platform are: a. Drastic time reduction: Decreasing the drafting time of registry entries from approximately 5 minutes to less than 45 seconds. b. Process standardisation: Ensuring that all registrars apply uniform criteria, regardless of their registry zone or level of experience. c. Error minimisation: Reducing errors of omission through tools that identify relevant data that might be overlooked during manual reviews. d. Capacity-building: Training registry staff in advanced digital competencies through the Learning and Study Centre. e. Citizen service improvement: Faster registrations without compromising technical quality, clearer and more complete observation notices (esquelas), and more complete responses.
Challenges and lessons learned
Implementation challenges: a. Resistance to change: Overcoming initial distrust of disruptive technologies through voluntary and collaborative adoption strategies. b. Training over 1,000 public servants across 14 registry zones with varying levels of technological familiarity. c. Effective prompt design: Developing methodologies so staff learn how to "converse" with the AI to obtain technically accurate responses. d. Template validation: Ensuring that entry (asiento) and observation (esquela) templates maintain the required legal precision. e. Technological integration: Co-ordinating the development and implementation of APIs and systems with the Information Technology Office. Lessons learned: a. Voluntary adoption works better than top-down imposition: Early adopters were enthusiastic volunteers who later became informal trainers for their colleagues.
