Justice at your fingertips: How AI is helping Delhi’s judges, lawyers deal with caseload

– In Delhi’s lower courts, there are nearly 15 lakh pending cases — 12.9 lakh criminal and 2.13 lakh civil, according to data available on the National Judicial Datagrid.

– The Delhi High Court has less than one judge per 10 lakh population, according to the India Justice Report 2025 by Tata Trusts, which analyses the criminal justice system.

– Close to three-quarters of the prisoners in India are undertrials, and the fate of over 50 million cases remains undecided. As things stand, it will take over 300 years to clear the current backlog of cases.

With courtrooms bogged down with manual processes and delays, is there a solution to this legal tangle?

Artificial Intelligence (AI) could provide one.

From AI-driven speech recognition software designed to assist court stenographers in transcribing witness examinations and orders dictated by judges to applications designed to streamline workflow, AI has been slowly transforming the judicial system in the country. While leaving the delivery of justice to the judges, it can automate all other processes that are too cumbersome for people.

Utkarsh Saxena, a lawyer since 2012, has co-founded one such firm that has come to the aid of the court. Called Adalat AI, it provides machine learning-powered solutions to courts in India in their attempt to tackle widespread backlogs and delays.

Saxena, who is also the firm’s CEO, said, “… Sometimes judges have to write their order manually… It’s a lot of paperwork. Some told us that they feel like their board exams are still going on.”

He used to be a clerk with a justice of the Supreme Court of India. “A large part of the problem of delay is caused not due to questions of law but those of logistics… our goal is to eliminate that. Tools can run in the background to quicken the slow processes, and at the same time, judges can focus on delivering justice,” he added.

With the Adalat AI application, all judges need to do is login, press two buttons, and speak out their observations. All their observations are automatically transcribed — legal jargon included — and voila! An order is automatically prepared.

While other speech-to-text applications exist, what separates Adalat AI from the rest is that it has been extensively trained on Indian legal jargon, Indian accents, and pronunciation. The advanced algorithm is calibrated using legal terms sourced from over 1 lakh court orders in 11 local languages.

The app also has a case flow management dashboard (analogous to a digital version of a cause list) that helps in streamlining workflows. All the judge needs to do is dictate his order, and it is converted automatically into the printed order. After the judge finishes dictating orders for the day, they can download them — all of which are converted into ZIP files.

The application is currently being used in 8 states and 3,000 courtrooms across the country — Kerala, Karnataka, Andhra Pradesh, Delhi, Bihar, Odisha, Haryana and Punjab. It aims to expand its reach to 50% of all courtrooms in the country by 2025.

“Adalat AI is also a part of the training curriculum at judicial academies,” Arghya Bhattacharya, co-founder of the firm, told The Indian Express. He is an AI and machine learning engineer.

“We have a tight feedback loop. What we’re building is for the system and also with the system. We have WhatsApp communities which enable judges or court staff to report any bugs in our application,” he added.

Adalat AI also allows scanning and converting legal documents into digital records instantly. “Our end goal is to liberate the justice system from manual clerical work,” said Bhattacharya.

At Delhi’s lower courts, the AI software is being rolled out. Staff seem to be relieved that, in the future, it could help reduce their workload. “Just look at these trunks,” said a court staff member working in a Delhi court, referring to the case files inside.

“Is there a need to have so many files? Half our time goes into sorting them out,” the staffer added. “I think files should be digitised to the maximum extent possible.”

Speaking at an event in July 2024 during which Adalat AI was inaugurated in Delhi’s district Courts, the then Delhi High Court Chief Justice — now SC Judge — Manmohan had said, “The biggest problem I see judges facing is that there is a large demand for stenographers, but there’s not a large pool available. I think this app will solve that problem to a large extent… it will ensure [by transcribing orders] that a large pool of stenographers will become available for other purposes.”

“Their team has done a fantastic job… this is the right application of technology,” he had added.

Along with judges, litigants, and stenographers, lawyers have also started relying on AI.

“I sometimes use an AI application to help me cull out important bits from judgments,” said advocate Vivek Chandra Jaiswal. “I just upload the PDFs and it gives me all the important portions. I’ve even used these arguments in court at times.”

One such application is ‘Ask Junior’, designed by advocate Renu Gupta. This AI tool concisely summarises judgments and court decisions. The tool also releases a monthly digest of summaries of Supreme Court judgments to aid advocates.

Gupta told The Indian Express, “We take a multi-step approach to generating our summaries. First, we parse each judgment and extract key components — facts, legal issues, arguments, reasoning, and the ratio decidendi. Then, we distill these into a short, accessible summary that captures the core legal reasoning without losing depth.”

“We’ve collected all publicly available Supreme Court judgments and used them to train our models, which are continuously evaluated using custom metrics developed with inputs from practicing lawyers,” she added.

Gupta also said that as of April 30, 2025, AskJunior had summarised 14,823 judgments with a 96.1% accuracy, delivered over 3,00,000 newsletters, and grown to 3,600+ Substack subscribers.

The Supreme Court of India has also adopted the use of AI language technology in the translation of judicial documents.

Till January 2025, 17 High Courts in the country have started providing online access to judgments in vernacular languages. On the same date, 36,324 Supreme Court judgments were translated into Hindi, and 42,765 judgments had been translated into 17 other regional languages.