The corpus.
Every cite back to the source.
Indian legal AI lives or dies on its corpus. Gotham reads the primary sources directly — the gazette, the judgment as the court released it, the circular as the regulator issued it — so every answer carries a verifiable citation back to the original. No model hallucination, no second-hand summaries, no orphaned quotes.
By the numbers.
A snapshot of what is in the corpus today. Counts roll over nightly as the ingesters land new material from courts, tribunals and regulators.
[TODO: live counts from BE]
The courts.
From the Supreme Court down to district eCourts. Coverage figures are pulled directly from the court's own e-portals; nothing is scraped from secondary aggregators.
Every reported and unreported judgment, daily orders, constitution bench references, registry circulars and roster notifications. [TODO: live count] judgments indexed.
District and taluka-level orders from the eCourts portal. Phased rollout begins with commercial-court benches in Q3 2026, expanding to all civil and criminal benches by end-2026.
The tribunals.
Where most commercial, tax and regulatory matters actually get decided. Each tribunal connector talks to the tribunal's own portal — no third-party aggregators.
National Company Law Tribunal — all benches, daily orders + final judgments since constitution in 2016.
National Company Law Appellate Tribunal — Delhi and Chennai benches, full text since inception.
Income Tax Appellate Tribunal — 63 benches across India, orders since 2010, key earlier orders backfilled.
Customs, Excise & Service Tax Appellate Tribunal — all regional benches, refreshed daily.
National Consumer Disputes Redressal Commission — all consumer matters, state commission orders in pipeline.
Debt Recovery Tribunals + Appellate Tribunal — SARFAESI and RDDB Act matters, all benches.
Securities Appellate Tribunal — SEBI, IRDAI and PFRDA appeals, full corpus since 1997.
Armed Forces Tribunal — principal bench Delhi + regional benches, service-matter jurisprudence.
Regulators & the gazette.
A regulator-issued circular often matters more than the parent statute. Gotham pulls each regulator's output directly from its own portal, on the day it's published.
Reserve Bank of India — circulars, master directions, monetary policy, FEMA notifications, RBI orders.
Securities and Exchange Board of India — ICDR, LODR, PIT, AIF, FPI regulations; final orders + informal guidance.
Central Board of Direct Taxes — circulars, notifications, instructions, AAR rulings, departmental press releases.
Central Board of Indirect Taxes & Customs — GST circulars, customs notifications, drawback rules, AAR rulings.
Ministry of Corporate Affairs — Companies Act notifications, LLP rules, IBC rules, NFRA orders.
Insurance Regulatory and Development Authority — circulars, regulations, exposure drafts, final orders.
Telecom Regulatory Authority of India — tariff orders, regulations, consultation papers, recommendations.
Competition Commission of India — orders, combinations approvals, leniency decisions, market studies.
Extraordinary and weekly editions ingested daily within the publication window. The single source of truth for every central enactment, notification and rule.
Phase 2 — starting with Maharashtra, Karnataka, Tamil Nadu, Delhi NCT, Gujarat and West Bengal in late 2026.
Update cadence.
If the regulator published it today, you can cite it today. Stale legal data is worse than no legal data.
Provenance, not vibes.
Every record in the corpus carries the receipts. You can open any cited authority and audit exactly where it came from, when it was ingested, and whether the underlying source has since changed.
The canonical URL on the issuing portal — supremecourtofindia.nic.in, sebi.gov.in, egazette.gov.in — preserved on every record.
SHA-256 of the original PDF or HTML as fetched. If the source portal silently re-publishes a document, the hash mismatch flags it for re-ingest.
UTC timestamp of first ingest plus every subsequent re-fetch. You can prove the version of the law your matter relied on at any point in time.
Every citation surface in Gotham — brief, memo, redline, chronology — lets you click through to the source, the hash, and the ingest log. No black boxes.