Legal research software for Indian law, built on the whole record.
Two ways to search 315.6 million court records. Keyword for when you have the words, AI semantic search for when you only have the idea. Every result links back to the judgment at its official source.
Corpus last counted 19 September 2026. Sourced only from official government portals, with no reseller in the chain.
What it does
Six things you will actually use
Not a feature list written to fill a comparison grid. These are the parts of the product that change how a research task runs.
Keyword search across the whole corpus
Exact word matching across 315.6 million records, with filters for court, year, judge, case type and date. This is the mode for when you already have the words: a section, a party, a case number, a phrase from a judgment you half remember.
AI semantic search for when you only have the idea
Describe a legal issue in plain language and get cases back that never use your wording. Semantic indexing covers a roughly 2 million judgment subset rather than the full corpus, and the product says so where it matters rather than implying a model has read all of it.
Citation network
See how a judgment has been treated by later courts: followed, distinguished, overruled or referred. This is a map of how cases reference each other, derived from the record, not an editor's written opinion on whether a case is still good law.
Similar case finder and grounded case chat
Start from one judgment and find the cases that resemble it on issues and facts rather than on vocabulary. Ask questions of a case and get answers tied back to the text of the record, with the source on screen.
AI case analysis
Issues, arguments, statutes relied on, precedents, case strength, chronology and issue spotting, produced against a specific judgment or your own uploaded matter. Outputs are a research starting point for a lawyer to verify, not a substitute for reading the judgment.
Acts and provisions alongside the case law
Bare Act text and individual provisions sit alongside the judgments, so a statutory question and the cases applying that section are reachable from the same place instead of two different subscriptions.
Who it is for
Four kinds of user, one corpus
The common thread is that all four need more than reported appellate judgments.
Practising advocates and solo practitioners
The corpus goes well past reported appellate judgments into the district judiciary and tribunals, which is where most Indian litigation actually sits. If your practice is section 138 complaints, execution petitions or NCLT work, a reported judgments archive will be quiet on most of your files.
Litigation teams in law firms
Research runs alongside matter management and contract work under one login with pooled credits, so a precedent found during research lands in the matter it belongs to rather than in somebody's downloads folder.
In-house legal and compliance teams
Party level litigation history and tribunal coverage answer the questions in-house teams are actually asked, which are usually about exposure and counterparties rather than about doctrine.
Legal researchers and academics
Filters on court, year, judge and case type make it possible to construct a defined set of judgments rather than a pile of search results, and judgment level metadata is exposed rather than buried.
The difference
A docket corpus, not a reported judgments archive
Most Indian legal research products are built on reported judgments: decided, usually appellate, usually edited opinions. That is a genuinely valuable corpus and the publishers who built it are better at it than we are.
It is also a small slice of Indian litigation. CourtMesh indexes the record as the registries publish it, which means District Court filings, live case status, order sheet timelines and party level litigation history across the Supreme Court, all 25 High Courts, the district judiciary and tribunals. If your question is what has been held on a point of law, a reporting publisher answers it well. If your question is what is pending, against whom, and when it is next listed, a reporting publisher cannot answer it at all.
The second difference is that both the corpus figure and its limits are published. The count above carries the date it was last taken, the split between keyword reach and AI analysed subset is stated rather than blurred into one number, and the coverage breakdown by court and year is on the coverage page instead of in a sales deck.
Honest limits
What this will not do for you
If one of these is central to your work, buy something else for it, or buy both. We would rather you knew now.
- No editorially written headnotes or catchwords. The established Indian reporting publishers have decades of that work and CourtMesh has none of it.
- No curated citator. The citation network shows how cases reference each other; it will not tell you authoritatively that a judgment has been overruled.
- No journals, commentary or secondary literature.
- AI semantic indexing covers a roughly 2 million judgment subset, not the full corpus. A record can be fully retrievable by keyword and carry no semantic indexing at all.
- Coverage is not uniform. District Court records carry less metadata than High Court records, and older records are thinner than recent ones across every court.
- This is a view of what registries published, not the official record. Confirm any position against the record of the court concerned before you rely on it.
Related
Where to go next
Plans and pricing
What each tier includes, published rather than quoted on a call.
The same corpus as an API
REST endpoints with published per-call pricing, for building on rather than searching in.
How we compare, with sources
A sourced comparison of Indian legal research platforms, including where competitors are genuinely stronger.
More writing on Indian legal research method is on the blog.
FAQ
Legal research questions
Put ten of your own questions to it
No demo survives contact with real research questions off real files. Take ten from your own matters, including the awkward District Court one, and see what comes back.