As an intern at the Supreme Court of Nepal, I had the opportunity to observe the daily life of the court up close. I watched how files moved, how cause lists were prepared, how summonses and dates were handled, and how court employees worked between old manual practices and newer digital systems. These observations made one thing clear: Nepal’s judiciary has entered the digital age, but it has not yet entered the age of artificial intelligence.

Public conversation around court technology is often too loose. Online cause lists, scanned records, SMS notices, virtual hearings, digital summons, automated templates and artificial intelligence are sometimes discussed as if they belong to the same stage of reform. They do not. A scanned record is not automation. A video hearing is not artificial intelligence. A template that fills in a reference number is not machine learning. If Nepal wants a serious debate on technology and justice, it must first be honest about where the system actually stands.

Movement of Nepal’s judiciary has been gradual, uneven and often pushed by necessity. Planned judicial reform began with the First Five-Year Strategic Plan in 2004, when court work was still closely tied to physical files and handwritten records. Later, the Long-Term Information Technology Master Plan, 2072–2082, gave clearer direction to court networking, automation and e-services. The Fourth and Fifth Five-Year Strategic Plans continued this path through IT-based case management, differentiated case management, technology-friendly services and the goal of a paperless court.

The SC introduced automated case management in 2023 to make case listing more transparent and reduce concerns over manual influence. Chief Justice Dr Manoj Kumar Sharma’s reform agenda has pushed the discussion further, including e-court, virtual hearings, electronic case files, live court proceedings, audio-to-text systems and feasibility studies on artificial intelligence, blockchain and machine learning.

Digitalisation, automation or AI?

Digitalisation means making court information and services available through digital systems. When cause lists are uploaded online, official summonses can be viewed digitally, old records are scanned, or case information is available on a website, the system becomes more accessible.

Automation goes a step further. It begins when a system performs repeated tasks by following fixed rules. If a summons template automatically pulls a reference number, if SMS alerts are sent for registration, peshi or tarekh, or if routine notices are generated through a case management system, that is automation. It reduces manual repetition, but it does not think, reason, or judge.

AI is different. It would mean a system that can analyse records, assist legal research, transcribe hearings, summarise documents, translate court materials, or identify patterns under human supervision. In Nepal, this remains more of a future discussion than a working reality.

Digitised, but not yet intelligent

The SC website provides public application formats, and writ registration timelines have been tightened at the Court. Online case registration for writ petitions has also been discussed around the court’s e-filing direction. These steps improve access, but they remain administrative reforms.

The internal court structure also shows why full automation is difficult. Even where a digital system exists, the work still moves through defined roles such as the registration or filing counter, the court bench and the concerned department or section, including section chiefs and dealing clerks.

NID-linked court service remains a future-facing possibility, not a completed judicial AI system. Nepal’s wider NID implementation itself has faced legal and administrative delays, so any court-level use must be handled carefully with privacy, legality and reliability in mind.

Nepal’s judiciary has digitised parts of its public service and is moving towards better administrative automation. It has not yet reached artificial intelligence in any serious institutional sense.

Harder test of court technology

The harder part of digital reform begins after the first tools become visible. A cause list on a website, a public CMS, online forms or scanned records can make court services easier, but they do not by themselves make the system secure, complete, or crisis-proof. The damage to the SC during the Gen Z movement exposed a serious weakness in judicial data protection and continuity planning.

For a judiciary speaking of paperless courts, this was not only an incident of damage. It was a warning about record protection, backup systems and institutional continuity. It must also mean protecting records when buildings are damaged, servers fail, or files are lost. If a court has to recover and authenticate destroyed case files after a crisis, record recovery itself becomes part of justice.

For many litigants, a lawyer’s physical visit to court still feels like proof that work is being done. For many lawyers, the visible effort of going to court, asking for dates, following up on summons or checking files has traditionally helped justify professional service, even when the real value should be measured by accuracy, strategy and outcome. Technology does not only change filing systems; it changes the culture through which legal work is trusted, billed and explained.

This is also why Nepal must be careful with AI. AI tools need reliable data, clean records and strong institutional control. Singapore can issue formal guidance on the use of generative AI tools by court users because its debate has already moved towards responsible use of AI in court materials. Nepal’s concern is still more basic: keeping files safe, making digital services consistent and ensuring that technology supports justice even during disruption.

Nepal’s judiciary does not need to prove its progress by using the word AI too early. Online registration, reliable summons tracking, SMS updates, secure case management, digital access to orders and better record-sharing between courts can do more for ordinary justice than a premature AI announcement.