Legal Tech Glossary
Plain-language definitions of the AI, legal-tech, and Indian court-process terms that come up across this site - written for a practicing advocate, not an engineer. This is general reference information, not legal advice.
RAG (Retrieval-Augmented Generation)
An AI approach where a model's answer is generated from specific retrieved documents or passages, rather than purely from what it learned during training. In legal use, this means an answer can be traced back to an actual document instead of being an unattributed guess.
AI Hallucination
When an AI system generates a confident, fluent-sounding statement - a case name, a citation, a fact - that is false or doesn't exist. In legal research and drafting, this most often shows up as a citation to a case that was never decided, or a real case being made to say something it doesn't.
Citation Verification
The process of confirming that a cited case or provision actually exists, is correctly attributed (court, date, case number), and genuinely supports the proposition it's cited for - as opposed to accepting a citation at face value because it reads plausibly.
Neutral Citation
A case reference assigned by the court itself (format: YYYY INSC N for the Supreme Court of India), independent of any commercial law reporter's own volume/page numbering.
CNR (Case Number Record)
A unique 16-character identifier assigned to every case filed in Indian courts under the eCourts system, used to look up a case's status, hearing history, and orders regardless of which specific court or state portal it was filed in.
eCourts
The Indian judiciary's national e-governance initiative for digitizing court processes - case status, cause lists, orders, and judgments - across district and subordinate courts, run under the eCommittee of the Supreme Court of India.
Cause List
The official daily list published by a court naming the cases scheduled to be heard that day, in order, along with the courtroom and presiding judge or bench.
Matter Management
The practice-management discipline of tracking a legal matter's full lifecycle - parties, filings, hearings, deadlines, documents, and communications - in one continuous record rather than scattered across separate tools.
OCR (Optical Character Recognition)
Technology that converts scanned or photographed document images into machine-readable, searchable text - a prerequisite for a system to be able to search or extract facts from a scanned court order or filed document.
Document Intelligence
AI-assisted extraction of structured facts, dates, parties, or key clauses from unstructured documents (orders, contracts, pleadings), used to populate a case record or answer a question without an advocate manually re-reading the entire document.
Source-Grounded AI
An AI system whose output is tied to an identifiable, checkable source (a specific document, passage, or judgment) rather than generated purely from a model's general training data with no attached reference.
AI Governance
The set of policies, review processes, and controls an organization puts in place to manage how AI tools are selected, used, and monitored - covering data handling, output verification, and accountability for AI-assisted work.
Subprocessor
A third-party service provider that a company relies on to process data on its behalf - for example, a cloud hosting provider, an AI model provider, or an email delivery service - disclosed so customers know who else touches their data.
Data Fiduciary / Data Processor (DPDP Act)
Under India's DPDP Act, 2023, a Data Fiduciary determines the purpose and means of processing personal data and bears primary compliance responsibility; a Data Processor processes data on the Fiduciary's behalf and instructions. A law firm using a legal-tech tool is typically the Fiduciary, with the tool acting as Processor - though the specific allocation depends on the actual contract.