The operating system for Indian law is the layer that turns India’s scattered public court records into one dependable source that people, companies and AI models can build on. It has three layers: a data infrastructure layer that ingests, cleans and links every court record; a portal and app layer where lawyers and litigants search, track and read cases; and an application and AI layer where tools like an AI clerk, background checks and legal agents run on top. All three sit on public, government-operated court rails. This page defines the category, shows what each layer does, and maps each one to a product you can use today on eCourtsIndia.
Last updated: 23 September 2026
Key takeaways
- India built public data utilities for payments (UPI), identity (Aadhaar), tax (GSTN), tolls (FASTag) and share ownership (NSDL and CDSL). Law is the large industry still missing its data spine.
- The operating system has three layers: data infrastructure, portal and apps, and applications and AI. Each layer only works because the one below it exists.
- On eCourtsIndia the data layer holds 32 crore+ case records and 125 crore+ orders and judgments from the Supreme Court, all 25 High Courts, district and taluka courts in all 36 states and union territories, and 18 tribunal and commission types.
- Every layer maps to a live product: free case search, cause lists and directories, the AI Clerk, LegalCheck, IndiaCode, the API and the MCP server.
- Search stays free. We charge for the convenience layers on top: AI, tracking, alerts and bulk access.
India already built five data utilities. Law is the sixth.
In fifteen years India built a run of national data layers. Banking got UPI on top of NPCI. Identity got Aadhaar. Indirect tax got GSTN. Highway tolling got FASTag. Share ownership got the demat stack on NSDL and CDSL. Each one took an activity that was already happening, and already producing data, and gave it one standard layer that applications could trust. The applications came second. The productivity gains kept compounding for years after.
UPI is the cleanest example. Bank transfers were happening long before 2016, just on dozens of incompatible rails. NPCI built one layer of standardisation, and within a few years every fintech app, every kirana QR sticker and most household transfers ran on it. The apps did not invent the activity. They removed the friction in the data underneath.
Law has not had that moment. More than 5 crore cases are pending across Indian courts according to the National Judicial Data Grid, which refreshes the count daily. Those matters touch litigants, lenders, employers, insurers and counterparties across the economy. The Bar Council of India told the Supreme Court in 2023 that about 25.70 lakh advocates were enrolled. Yet the records they all depend on sit across the Supreme Court site, 25 High Court portals, the district eCourts services and a long list of tribunal sites that do not search across each other and were never built for cross-court work.
What an operating system actually is
In software, an operating system is the layer between raw hardware and application code. It takes the messy, varied reality of the machine and exposes a small set of clean, dependable interfaces. An app does not need to know which disk controller is installed. It calls a file API and trusts the OS to translate. That abstraction is what makes a thousand applications possible.
An operating system for Indian law applies the same idea to court data. The hardware, in this analogy, is the Indian judiciary: the Supreme Court, 25 High Courts, district and taluka courts, tribunals and commissions, plus the registries, cause lists and orders they produce every day. These are public rails run by the Indian state. They generate an enormous amount of legally significant data, but they expose it in many formats, across many sites, with no shared schema and no common API. A developer or a lawyer who wants to work across that reality needs something in between.
So an operating system is not a website and it is not a search box. It is the layer that decides what every application above it can do. For Indian law it needs four things at once: a single structured view of every court, order, party, advocate and judge, refreshed daily and linked across portals; an interface that practising advocates will use every day, including on a phone; developer access so banks, background-check firms, insurers and founders can build their own products; and an agent layer that lets large language models read the corpus natively.
The stack at a glance
The diagram that used to sit here was a picture of three stacked layers. Here is the same picture as a table you can read, copy and cite.
| Layer | What it does | Live on eCourtsIndia today | Role |
|---|---|---|---|
| Layer 3: applications and AI | AI clerk, background checks, due diligence, litigation analytics, legal agents | AI Clerk, LegalCheck, IndiaCode, MCP server | The most visible layer. Built by many companies, least defensible on its own. |
| Layer 2: portal and apps | Unified search, cause lists, lawyer, judge and litigant pages, alerts | Case search, cause lists, verified advocates, Android, iOS and Chrome apps | The distribution engine. Where people build daily habits. |
| Layer 1: data infrastructure | Ingest, clean, de-duplicate, resolve entities, enrich, index, serve | 32 crore+ case records, 125 crore+ orders and judgments, API with 23 endpoints | The foundation. Slow to build, compounds every day. |
| Upstream public rails | Official court systems that create the record | ecourts.gov.in, NJDG, Supreme Court and High Court portals, tribunal sites | Public and government-operated. Everything above depends on it. |
Layer 1: the data infrastructure
The foundation is the data layer, and it does seven jobs. It ingests from public upstream sources such as ecourts.gov.in, NJDG, the High Court sites, the Supreme Court portal and tribunal sites. It cleans inconsistent formats. It de-duplicates records that appear in more than one place. It resolves entities so that one advocate with five spellings of their name becomes one record. It enriches the raw data with summaries, tags and links. It indexes everything for fast search. And it serves the result through a consistent, documented API.
On eCourtsIndia that layer now covers 32 crore+ case records and 125 crore+ orders and judgments. It links them to 34 lakh+ advocate names, 82,000+ judges and 79 crore+ litigant records, and it carries 7 crore+ cause-list entries. Every record keeps its CNR, the 16-character identifier that survives transfers when case numbers do not. Our guide to mapping India’s court data stack walks through where each piece comes from.
Why the data layer is the hardest part
Anyone can wrap a prompt around a language model. Very few teams can build the data layer underneath it. Indian court data is difficult in ways that do not exist in most Western systems. The 25 High Courts publish in their own formats. District establishments publish in regional variants. Orders arrive as scanned PDFs in Hindi, Marathi, Tamil, Telugu, Bengali, Kannada, Gujarati, Odia and Punjabi. Cause lists change overnight. A party is spelt three ways in three filings, and the same advocate shows up under three name variants across three benches.
Building the structured layer means solving all of that at once: harvesting from every source, running OCR on scanned vernacular orders, normalising schemas across courts, linking each order to its case, bench, judge, advocate and party, and keeping everything fresh when one portal renames a field on a Tuesday morning. You cannot scrape your way to that. It takes a pipeline that treats every court as a first-class source and every record as a permanent object. We wrote about the engineering side in From PDFs to APIs.
That pipeline is not glamorous, but it is the moat. Once it exists, applications above it can be built in weeks. Without it they cannot be built at all, which is a large part of why Indian legaltech spent two decades on Supreme Court and High Court judgments and left the district tier, where most litigation lives, largely unserved. We look at that gap in why legaltech keeps missing the district court lawyer.
Layer 2: the portal and apps
The middle layer is where people meet the data. Working lawyers, litigants, researchers, journalists and compliance teams use it to find a case, follow a courtroom, check a counterparty or read an order. On eCourtsIndia that layer includes:
- Unified case search across every court level by party name, CNR, advocate, judge, case type and status, free and without a login, at ecourtsindia.com/search.
- Daily cause lists for district and taluka courts at ecourtsindia.com/causelist.
- Lawyer, judge and litigant pages built from court records, plus a separate directory of bar-card verified advocates that litigants can contact directly.
- Order PDFs and AI order summaries, charged at Rs 2 each in credits.
- WhatsApp and email case alerts when a tracked case gets a new hearing date, order or status.
- Crime Reports, a search over 12 lakh+ FIR PDFs from 13 states and union territories at ecourtsindia.com/CrimeReports.
- Add a Missing Case at ecourtsindia.com/add-case, so a reader who cannot find a matter can ask for it to be pulled in.
- Apps for Android and iOS, and a Chrome extension that shows tomorrow’s listings and case updates inside the browser.
The portal is the distribution engine. It is how people discover the operating system, and how they build the daily habits that later turn into paid use of the layer above. It is also where the public rails stay free: searching, cause lists and the directories cost nothing.
Layer 3: applications and AI
The top layer is where the visible products live. Some are ours, and many more will be built by others on the same data. Five are live today.
AI Clerk
The AI Clerk is a working assistant for advocates and litigants. It tracks cases, reads orders, answers questions against the record and sends alerts. It runs on credits that never expire: a free plan with 50 credits a month, Rs 250 + GST a month for 500 credits, Rs 5,000 + GST for six months (7,500 credits) or Rs 10,000 + GST a year (20,000 credits). Case tracking is Rs 5 per case per month and each WhatsApp or email alert is Rs 0.50. The full table is on the pricing page.
LegalCheck
LegalCheck is an identity-first legal background check for people and businesses. It turns the same court corpus into a report a hiring, lending or onboarding team can act on, and it is also available to partners as an API at Rs 99 per check pay-as-you-go, or Rs 33 with a subscription. The LegalCheck launch post explains how a report is built.
IndiaCode
IndiaCode by eCourtsIndia is the statute side of the stack: 10,084 Acts and 2,69,602 provisions, each section linked to 4,900+ reported judgments that read it. A court record tells you what a court did. IndiaCode tells you what the law says. Our IndiaCode guide shows how the two connect.
The API and the MCP server
The eCourtsIndia API exposes the data layer to builders through 23 endpoints, with Rs 200 in free credits on signup and documentation at ecourtsindia.com/api/docs. The MCP server at https://mcp.ecourtsindia.com/mcp goes a step further. It lets AI assistants such as Claude call 39 tools across cases, cause lists, the statute book and the electoral roll, so the model can check the record at the moment it reasons instead of guessing. The more agents depend on that substrate, the deeper it sits under the category. Start with how to connect the eCourtsIndia MCP to Claude, and read the AI agent layer for Indian law for the bigger picture.
An individual application is easier to build than the layer below it, and also easier to copy. Applications with direct access to a strong data layer inherit its defensibility. Applications without it rent that strength from someone else.
Who the operating system serves: three buyer pools, one spine
The same asset serves three structurally different groups without splitting the engineering team.
Practising lawyers and litigants are the largest group. Advocates need unified search, cause lists and a clerk that watches their matters. Litigants need to find their own case, understand an order and know the next date. The price has to start at zero and grow with the work, which is why search is free and the AI Clerk starts on a free plan. Surveys already show that many legal professionals use AI. We look at what that means in why data, not adoption, is the bottleneck for Indian lawyers using AI.
Enterprises are the most valuable per account. Banks and NBFCs check borrower litigation before they disburse. Background-verification firms replace manual court checks with an API call. Insurers look for parallel claims. Corporate legal teams monitor their portfolio. Journalists and researchers use the same data to study how courts work. An API here replaces an existing budget line such as paralegal time or vendor checks, rather than asking for new money. The case for this is in why India needs a CIBIL for litigation.
Developers and AI agents are the long tail. Every legal AI startup in India needs court data to be useful. The default path is to scrape, fail at scale and lose quarters to data plumbing. The shortcut is an API. The accelerator is an MCP server that lets the agent compose the workflow itself.
Each pool funds the spine, and the spine improves for each pool. Lawyer usage keeps case files fresh, which improves enterprise outputs. Enterprise revenue pays for vernacular OCR, which makes the lawyer product work in Kannada and Hindi district courts. Agent traffic surfaces edge cases that improve the schema for everyone. They are not three bets. They are three ways to use one asset.
Why this framing matters: feature mindset versus OS mindset
| Question | Feature mindset | Operating system mindset |
|---|---|---|
| What is the product? | A drafting tool | A data and interface layer that enables many products |
| What is the moat? | Better prompts, better UI | A multi-year data pipeline that compounds |
| Who is the customer? | Individual lawyers | Lawyers, litigants, enterprises and other legaltech builders |
| What is the failure mode? | A bigger competitor copies the feature | The upstream public data changes format |
The difference is not cosmetic. A feature is priced by what one user will pay for one task. An operating system is valued by every application built on top of it. One of those ceilings is small. The other is very large.
The public rails underneath
It is worth saying plainly. The upstream public systems, ecourts.gov.in, the NJDG and the individual court portals, are and should remain government-operated. The eCourts Mission Mode Project, run by the Supreme Court’s e-Committee and the Department of Justice, has invested INR 935 crore in Phase I and INR 1,670 crore in Phase II, and Phase III carries an outlay of INR 7,210 crore to produce and serve this public data at scale. Our Phase III breakdown shows where that money goes.
A private operating system sits on top of those rails the way PhonePe and Google Pay sit on UPI, or the way Zerodha sits on NSE and BSE. The exchanges run the rails. The broker builds the interface that makes them usable. The public layer provides the data and its legal legitimacy. The private layer adds experience, reliability, speed and integration. NPCI does not build consumer payment apps, and the eCourts portal, a public records system that shows one case at a time behind a captcha, was never meant to be a product for cross-court work. That complementary slot is the one we believe in, and we set out the principle in Public Data, Private Experience.
Why a foreign player will not build this for India
Global legal AI has proved the category is large. Harvey raised USD 550 million at a USD 15.5 billion valuation in September 2026, with annual recurring revenue reported above USD 400 million. Legora raised a USD 550 million Series D at USD 5.55 billion in March 2026 and extended the round to USD 600 million at USD 5.6 billion in April. Thomson Reuters bought Casetext for USD 650 million in 2023. These are strong companies.
They are also built for a different market. Their corpora are English-language, common-law research and contract review, centred on American and British practice. They do not run Hindi OCR on district court orders. They cannot pull a CNR, and they hold no structured data from a Tier 2 district court in Maharashtra. Building India coverage from a US base is expensive and slow, so the winning operating system for Indian law will be local by necessity: local courts, local procedure, local languages and local price points. We compare the global players in more depth in India’s legal AI category and its global comparables.
What we will not do
An authoritative platform is defined as much by what it refuses to do as by what it ships. A few commitments you can hold us to:
- We will not publish numbers we cannot verify against a source.
- We will not let an AI agent invent a precedent that does not exist in the record.
- We will keep search, cause lists and the directories free. We charge for convenience layers such as AI, tracking, alerts and bulk access.
- We will not impersonate an official judicial system or present ourselves as a government portal.
- We will not fabricate quotes from judges or public figures.
These are the floor, not the ceiling. Trust in legal data is the most important thing we can earn from the people who rely on it.
What this means for eCourtsIndia
We build the data layer, the portal and apps, and the AI and developer surfaces on top, all sitting on the public foundation. The data layer covers the Supreme Court, all 25 High Courts, district and taluka courts in all 36 states and union territories, and 18 tribunal and commission types, and it keeps growing. The portal serves lawyers, litigants and researchers every day. The AI Clerk, LegalCheck, IndiaCode, the API and the MCP server expose the same spine to people, enterprises and agents. The goal is not to win any one application. It is to make the whole ecosystem possible, and to let many other teams build on it too.
The quickest way to see the operating system at work is to use it. Search a case, open the cause list for a court you appear in, or connect the MCP server to your AI assistant and ask it about a matter.
Related reading
- What Happens When Someone Owns the Data Layer: CIBIL, Zerodha, PitchBook, Clio, Bloomberg
- From Scrape to SLA: Moving Off services.ecourts.gov.in to the eCourtsIndia API
- Public Data, Private Experience: A Manifesto
- LegalCheck API: court-record BGV for India
Sources
- eCourtsIndia index counts (case records, orders, advocates, judges, litigants, cause-list entries), verified 23 September 2026.
- National Judicial Data Grid (njdg.ecourts.gov.in), pendency dashboard.
- Supreme Court of India, Ajay Shankar Srivastava v. Bar Council of India, order dated 10 April 2023 (about 25.70 lakh enrolled advocates).
- Department of Justice and Press Information Bureau, eCourts Mission Mode Project Phase I, II and III outlays.
- TechCrunch, “Harvey hits $15.5B valuation”, 9 September 2026.
- TechCrunch, Legora USD 5.55 billion Series D, 10 March 2026; The Global Legal Post, Series D extension to USD 5.6 billion, April 2026.
- Thomson Reuters acquisition of Casetext, 2023, USD 650 million.
Frequently Asked Questions
What is the operating system for Indian law?
It is the layer that turns scattered public court records into one dependable source that people, companies and AI models can build on. The bottom layer ingests, cleans and links records, the middle layer is a portal and apps for search and tracking, and the top layer holds AI tools such as an AI clerk and background checks. You can see it working at eCourtsIndia search.
What does the data infrastructure layer do?
The data layer does seven jobs. It ingests from public court sources, cleans inconsistent formats, de-duplicates records, resolves parties, advocates and judges into single entities, enriches records with summaries and tags, indexes everything for fast search, and serves it through a documented API. Builders can query it directly through the eCourtsIndia API, which has 23 endpoints and Rs 200 in free credits.
Which eCourtsIndia products map to each layer?
The data layer is the index of 32 crore+ case records and 125 crore+ orders, exposed through the API. The portal layer is free case search, cause lists, directories, the mobile apps and the Chrome extension. The application layer is the AI Clerk, LegalCheck background checks, IndiaCode for statutes and the MCP server for AI assistants. Plans for the paid parts are on the pricing page.
How is this different from the government eCourts portal?
The eCourts portal is a public records system that shows one case at a time, with a schema that varies by court. It was never built for cross-court search, bulk access or AI agents. The operating system sits on top of it as a private experience layer and does not replace it. We explain the principle in Public Data, Private Experience.
How much has the government invested in eCourts?
The eCourts Mission Mode Project has invested INR 935 crore in Phase I and INR 1,670 crore in Phase II, and Phase III carries an outlay of INR 7,210 crore to produce and serve public court data at scale. The public rails stay government-operated. We break down where the Phase III money goes in this detailed guide.
How many courts and records does eCourtsIndia cover?
eCourtsIndia covers 32 crore+ case records from the Supreme Court, all 25 High Courts, district and taluka courts in all 36 states and union territories, and 18 tribunal and commission types. It also links 125 crore+ orders and judgments, 34 lakh+ advocate names and 82,000+ judges. You can search any of them for free and open public judge pages.