⚡TL;DR
Adalat AI is a legal-tech nonprofit dedicated to transforming justice in developing countries through AI solutions that reduce court delays and improve judicial quality.
📜 Meet justice’s AI co‑pilot. Incubated at MIT/Oxford, Adalat AI builds an AI‑powered operating system for courts. It handles real‑time transcription in 15 + Indian languages with over 90 % accuracy and already serves over 2,000 courts and 3,000 judges.
🚨 50M cases & century‑long waits. India’s judiciary is swamped with pending cases; average resolution takes 13 years, and the backlog would take 300 years to clear. One of the primary issues is lack of transcription options as only 10–20 % of courtrooms have stenographers.
⚙️ AI scribes to streamline caseloads. The platform integrates real‑time speech‑to‑text, case‑flow management, document scanning and translation, chatbots and legal summarization to eliminate clerical bottlenecks and cut case timelines by 30–50 %.
🙌 Why We Say Hell Yeah. By speeding up hearings, Adalat AI helps keep undertrial prisoners out of jail and showcases one of the best use-cases for AI we’ve seen yet.
When the justice process itself becomes the punishment
India’s court system is supremely back-logged. This is more than an administrative headache as 280,000 people are held in jail during or awaiting trial, without having been convicted of a crime. According to Amnesty International, these prisoners known as undertrials, account for two out of three prisoners in India’s prisons - a percentage far higher than other democracies around the world.
Many have been awaiting trail for years, some for a longer period than their maximum formal sentence.
That’s crazy. Imagine what this does to communities while loved ones are imprisoned for years simply waiting to get a trial. This also has significant economic and societal impact as it traps daily wage working families in cycles of poverty (frequent court appearances cause lost earnings and job instability).
Note: Pre-trial detention is not only an Indian problem. In the US, more than 400K people are currently in jail awaiting trial. Many of these people may be released same-day, or when they post bond. A big difference is that in the US, the court systems move (relatively) quickly and the “pretrial” time in jail is smaller than India. Still though, in Illinois, the average pretrial time in jail is between 25 - 34 days.
More writing time, more jail time
There are a number of reasons why India’s legal system moves slowly, but one of the most pervasive and frustratingly well-known issues is transcription. By law, all courtroom comments must be precisely recorded - and there are structural issues that make this challenging.
Shortage of stenographers: Only 10-20% of Indian courtrooms have stenographers (mostly in urban areas). Judges (often in rural areas) write by hand, slowing proceedings, diverting them from adjudication duties, and creating general exhaustion.
Language barriers: India’s courts must accommodate over 140 dialects. Testimony must be translated into formal English for official record.
Paper‑based processes: Many courts still rely on physical records. Manual scanning and data entry cause delays and errors.
Some of the common dialects in India. Image taken from https://medium.com/@roshanabraham_32060/language-breaking-unity-in-diversity-83510793388a
The gavel meets the graph
Adalat.ai’s solution is a full-stack judicial operating system. The platform uses automatic speech recognition (ASR) models fine‑tuned on legal statutes and case law. It provides real‑time transcripts with 90 %+ accuracy and supports 15+ languages so far. The technology handles legal jargon and vernacular speech, meaning judges and stenographers no longer need to transcribe manually. Judges can edit and finalize transcripts within the system, ensuring accurate records.
Additionally, the company has document digitation, AI-powered research, chatbots, and litigant engagement tools that help address other specific problems in the system.
The team is also incredibly impressive. CEO Utkarsh Saxena is Harvard Law School educated former lawyer and law clerk on the Supreme Court of India. CTO Arghya Bhattacharya is an IIT educated leading AI and machine learning engineer. Their advisory board includes a number of retired lawyers and judges. As a nonprofit entity, they know what they’re doing and are going about it in a thoughtful way.
Impact chart from the Adalat.ai website
⚡Why we said hell yeah!
A legal system shouldn’t be held hostage by keystrokes. When a human being’s freedom hangs on a transcript, every delay is a quiet (and avoidable) injustice. I would be boiling if I were in the courtroom watching a judge slowly handwrite things while a loved one sat in jail awaiting trial. On a similar note - thank goodness I’m not a judge given my glacially slow and indecipherable handwriting.
This is AI at its best. Take the slowest, most tedious work in the room and make it disappear, so judges can judge, lawyers can advocate, and people can get on with their lives. We love this idea as its not an AI-gadget hunting for a use-case. It’s a screaming, frustrating, humanitarian problem that just needs a simple, trustworthy, and elegant solution.
Do it as a nonprofit, and the tech spreads where it’s needed most - beyond big-city budgets, into small district courts, into regions the market would price out. Start in India, then the Global South: multilingual, low-resource, with real-world trials that improve accuracy over time. And if we set the bar of AI-speech-transcription as the norm, we help open up other legal-tech solutions to help proceedings move faster. It can reset how justice moves everywhere.
Hell Yeah!
Dive in Deeper
MIT Solve - Adalat AI pitch
Draper Richards Kaplan Foundation - Overview of model and vision
Fortune Business Insights - Info on the growth of the global legal tech market
Verified Market Reports - Info on growth of AI transcription software
Prison Policy Initiative - Research on criminal justice statistics and policy, including in the US
Hit us up if you’d like to learn more or if you have suggestions for future features.
Forwarded this email? Subscribe at hellyeah.xyz
Don’t forget: life’s too short to be an Eeyore.
