Resources · Glossary

Glossary

Definitions of the litigation, legal AI, and platform terminology used throughout Discover Docket and across the litigation technology landscape. California and Federal procedural references where they apply. Discover Docket supports all 52 U.S. jurisdictions at launch — the focused references below reflect what most of our launch customers practice in.

This glossary is provided for informational purposes only and is NOT to be construed as legal advice. Definitions are general; application to specific matters requires the judgment of counsel.

Legal AI and ethics

Agentic AI: AI systems that can take actions in software environments — file documents, send communications, schedule events — beyond simply generating text. The California State Bar's May 2026 update to its AI Practical Guidance addresses agentic AI specifically.
Audit log: A chronological record of every action taken in a system. In the context of legal AI, an audit log should record every prompt, every retrieval, every output, and every confidence score, with cryptographic signing and chaining that makes the log tamper-evident. See DDEAS.
Confidence score: A numeric signal accompanying an AI output that reflects the strength of the underlying authority, the consistency of supporting sources, and the freshness of the data the output was drawn from. Confidence scores should be paragraph-level or claim-level, not whole-output level.
Citation validation: The process of confirming that a legal citation produced by an AI tool refers to an actual case, statute, rule, or other authority, and that the citation is current good law. In a defensible architecture, validation is a structural gate before the citation reaches the user — not a post-hoc verification step.
DDEAS: Discover Docket Ethical and Accountability Standards. The cryptographic and architectural framework that governs every AI output produced inside Discover Docket. Four layers: source validation, confidence scoring, cryptographic audit log, tamper-evident chain. See the DDEAS framework page.
Hallucination: Output produced by a generative AI model that appears authoritative but is factually false. In legal practice, hallucinations most commonly take the form of fabricated case citations — citations to cases that do not exist. Mata v. Avianca and Park v. Kim are the leading sanctions cases involving AI hallucinations. See the Mata v. Avianca article.
JILL: Judicial Insights & Legal Logic — the AI built into Discover Docket. JILL operates inside a litigation-specific data model and is constrained by the DDEAS framework. See the JILL page.
Large language model (LLM): A type of generative AI trained on large text corpora to produce coherent natural-language output. LLMs are the underlying technology behind most consumer AI chatbots and most bolt-on AI features in legal technology products. LLMs do not, by themselves, validate their own outputs against authoritative sources.
Mata v. Avianca: The canonical federal sanctions case on AI hallucinations. Decided June 22, 2023 in the Southern District of New York. Attorneys Steven Schwartz and Peter LoDuca were sanctioned $5,000 each for filing a brief with six fabricated case citations produced by a generative AI tool. See the article.
Park v. Kim: The Second Circuit appellate decision (January 30, 2024) referring Attorney Jae S. Lee to the Court's Grievance Panel for citing a ChatGPT-fabricated case in a reply brief. Federal Circuit-level authority on AI sanctions. See the article.
Retrieval-augmented generation (RAG): An AI architecture pattern in which the model retrieves relevant context from a separate database before generating output, rather than relying only on what it was trained on. RAG is a necessary but not sufficient component of defensible legal AI — the retrieval has to be from authoritative sources, the outputs have to be validated, and the audit trail has to be tamper-evident.
Source validation: Confirming that an AI-generated reference to authority (case, statute, rule, regulation) corresponds to an actual authority and is current good law. In a defensible architecture, source validation is a gate the AI's output must pass through before reaching the user.
Tamper-evident chain: A cryptographic structure in which each record's hash includes the previous record's hash, so that altering any historical record breaks a verifiable chain. The same architecture that secures financial transaction records and digital evidence chains of custody.

Litigation procedure (California and Federal)

Affirmation: In New York and certain other jurisdictions, a sworn written statement made by an attorney in lieu of a notarized affidavit, under penalty of perjury. The fabricated cases in Mata v. Avianca were submitted in an attorney affirmation.
California Code of Civil Procedure (CCP): The statutory framework governing civil procedure in California state courts. Cited in this guide and across Discover Docket as “CCP §” followed by section number.
California Rules of Court (CRC): The procedural rules adopted by the Judicial Council of California, supplementing the CCP. Cited as “CRC” followed by rule number.
Discovery cutoff: The deadline by which discovery must be completed. Under CCP § 2024.020, at least 30 days before the initial trial date in California. In federal court, set by the scheduling order under FRCP 16(b).
Federal Rules of Civil Procedure (FRCP): The procedural rules governing civil cases in U.S. federal district courts. Cited as “FRCP” followed by rule number.
Form interrogatories: Standardized written questions propounded in discovery, organized by category. In California, governed by CCP § 2030.020. The most commonly used form is Judicial Council Form DISC-001.
Local rules: The procedural rules of a specific court, layered on top of the statewide rules. In California, each of the 58 superior courts publishes local rules. In federal practice, each of the 94 federal districts publishes local rules. Local rules are binding on practice before the specific court.
Meet and confer: A pre-motion process in which counsel for the parties confer in good faith to attempt to resolve the dispute without court intervention. In California, required for most discovery motions under the Civil Discovery Act. Specific meet-and-confer requirements vary by motion type, by local rule, and by standing order.
Motion in limine: A pretrial motion asking the court to exclude certain evidence from trial. Procedure varies substantially by court and judge.
Motion to compel: A discovery motion asking the court to order a party to provide responses to discovery requests, or to provide further responses. Subject to meet-and-confer requirements.
Requests for admission (RFAs): Written requests asking a party to admit or deny specific facts. In California, governed by CCP §§ 2033.010 et seq.. Failure to respond timely results in the requests being deemed admitted — one of the most consequential procedural defaults in California civil practice. In federal court, governed by FRCP 36.
Rule 11 sanctions: Sanctions under Federal Rule of Civil Procedure 11 for filing pleadings, motions, or other papers that are frivolous, factually unsupported, or filed for an improper purpose. The sanctions in Mata v. Avianca were imposed under Rule 11.
Rule 37 sanctions: Sanctions under Federal Rule of Civil Procedure 37 for discovery violations. Includes dismissal as an available sanction for sustained noncompliance with discovery orders.
Service additions: Additional time added to a deadline when a paper is served by a method other than personal service. In California, under CCP § 1013: 5 court days for mail within California, 10 court days for mail outside California, 2 court days for fax or electronic service.
Special interrogatories: Custom written questions propounded in discovery (as distinguished from form interrogatories). Governed by CCP § 2030.030.
Standing order: A written directive issued by an individual judge governing practice before that judge or in that judge's department. Standing orders layer on top of local rules and statewide rules. Compliance with the statewide rule does not excuse noncompliance with the standing order.
Summary judgment: A pretrial procedure for resolving a case (or specific issues) without trial when there is no genuine dispute of material fact. In California, governed by CCP § 437c with a 75-day notice period. In federal court, governed by FRCP 56.

Platform terminology

Case dashboard: The matter-level view in Discover Docket that surfaces the deadline state, document state, recent activity, assigned team, and current work product for a specific matter.
Case type: A categorization of a matter by area of practice (medical malpractice, employment, business litigation, intellectual property, etc.). Case type determines which specialized workflows and substantive knowledge layers JILL applies.
Department: Within a California superior court, the specific courtroom and judge a matter is assigned to (e.g., “San Diego Superior Court Dept. 73”). Standing orders are typically department-specific.
Docket events: Significant events in the procedural life of a matter that trigger deadline calculations and team notifications. Includes filings, service, hearings, continuances, stipulations.
Ethical wall: A separation enforced between matter teams to prevent conflicts of interest. In Discover Docket, ethical walls are enforced at the database layer through row-level security, not at the application layer, so that a screened user encounters a structural barrier rather than a UI restriction.
Folder taxonomy: The litigation-shaped document organization structure Discover Docket uses for every matter: pleadings, discovery, depositions, expert materials, correspondence, motions, evidence, trial preparation. Different from generic flat-folder document storage.
Matter team: The set of users with access to a specific matter. Configured per matter; can be modified by the matter's lead attorney or firm administrator subject to ethical wall rules.
Practice area: A subject-matter category (civil litigation, family, criminal, employment, IP, real estate, probate, immigration). Determines which substantive law modules JILL applies.
Verification record: The cryptographically signed audit trail produced by DDEAS for a specific work session or output. Can be exported as a PDF and attached to the matter file, or produced on demand if the underlying work is challenged.

Security and compliance

AES-256: The encryption standard used for data at rest in Discover Docket. Industry-standard symmetric encryption used by financial institutions and government systems.
Application access policy: A policy in Microsoft 365 that restricts an application's ability to send mail or access mailboxes to a specific set of accounts. Used to lock down Mail.Send permissions to specific sender addresses.
Cryptographic signing: Attaching a cryptographic signature to a record that can be verified to confirm the record's authenticity and integrity. Used for every DDEAS audit log entry.
HIPAA: The Health Insurance Portability and Accountability Act, which governs the handling of protected health information (PHI). Discover Docket's architecture is designed to support HIPAA-aligned handling of medical records and other PHI in matters that involve them (medical malpractice, personal injury, workers' compensation, employment claims involving health records, family law matters touching health).
Hash chain: A sequence of cryptographic hashes in which each hash depends on the previous one. Used in DDEAS to make the audit log tamper-evident — altering any historical record breaks the chain.
No-cookies / no-tracking posture: Discover Docket does not use cookies, tracking pixels, third-party analytics that profile users, advertising trackers, or any other software that monitors visitor behavior across the web. Visitors to the marketing site are not tracked, and no information about them is sold, shared with advertisers, or shared with data brokers.
Row-level security (RLS): A database-layer access control mechanism in which each row of a table can be restricted to specific users or roles. Discover Docket uses RLS to enforce matter access, ethical walls, and firm boundaries at the database layer, not at the application layer.
SOC 2 Type II: An independent attestation of a service organization's information security controls under the AICPA Trust Services Criteria. Discover Docket is SOC 2 Type II compliant via enterprise-grade infrastructure. Customer data is stored, processed, and transmitted through environments under continuous independent SOC 2 Type II audit.
TLS 1.3: The current industry-standard protocol for encryption in transit. Used by Discover Docket for all client-server communication.
Zero-retention terms: A contractual commitment from an AI service provider that customer inputs and outputs will not be retained, used for training, or accessed by provider personnel. Required for confidentiality compliance under ABA Formal Opinion 512 and state bar AI guidance.

This glossary is provided for informational purposes only and is NOT to be construed as legal advice. Definitions are general; application to specific matters requires the judgment of counsel. Nothing in this glossary creates an attorney-client relationship between the reader and Chris Waters or Discover Docket.

The work gets done. You get to be the lawyer.

Join the waitlist. California firms first.