Jury Instructions Copilot for California Tenants

Jury Instructions Copilot for California Tenants

An AI copilot that helps tenants and legal aid teams generate, review, and finalize court-compliant jury instructions and trial documents for eviction cases in California.

Project Description

Preparing jury instructions for an eviction trial in California is a high-stakes task that many tenants and their legal teams struggle to complete correctly and on time. The process involves identifying applicable legal defenses, selecting the corresponding jury instructions from a 3,000+ page CACI template, inserting personalized case data, and formatting the final document correctly—all under tight deadlines and with limited resources.

This AI copilot is being developed by Stanford Legal Design Lab in collaboration with legal aid partners to address this challenge. It ingests key case documents (such as the UD-105 Answer and local LA CIV 244 form), extracts relevant defenses and claims, and uses this information to automatically assemble a customized set of jury instructions. It pulls from both the official California Judicial Council (CACI) templates and custom local instructions from a legal aid group.

The copilot personalizes the selected templates with case-specific information—names, pronouns, case numbers, trial dates—formats the instructions per court rules (e.g., one instruction per page), and presents the draft for user review. It supports both legal aid staff workflows and self-help scenarios where a tenant may be acting pro se (without a lawyer). The tool helps ensure compliance with court requirements and minimizes errors that could delay or derail a tenant’s trial.

Why This Matters

Jury instructions are essential for tenants to assert their right to a jury trial during eviction proceedings. If filed late or incorrectly, the court may delay or reject the tenant’s request, jeopardizing their defense. Housing attorneys are often overwhelmed and cannot support every tenant with this task, and help centers rarely have capacity to assist.

By automating the most tedious and error-prone parts of the jury instruction process, this copilot makes trial preparation more accessible, reduces clerical errors, and supports fairer outcomes in housing court.

AI Agent Capabilities

  • Document Ingestion: Accepts scanned or digital versions of UD-105 Answers and related materials.
  • Defense Identification: Uses NLP to extract legal defenses and key claims.
  • Form Logic Application: Automatically completes LA CIV 244 based on recognized claims and procedural needs.
  • Jury Instruction Assembly: Pulls and assembles CACI and LAFLA instructions relevant to the claims.
  • Personalization: Fills in party names, dates, and gendered language consistently throughout.
  • Formatting: Outputs jury instructions in court-compliant format (one per page, correct font/structure).
  • Review Mode: Allows legal staff or self-represented litigants to review, edit, and finalize.
  • Binder Preparation: Compiles all documents needed for a complete Trial Binder: Jury Instructions, Exhibit List, Witness List, Statement of the Case.

Data Sources

  • Judicial Council of California CACI Jury Instruction PDFs and HTML versions
  • Legal aid’s custom jury instruction templates
  • Court forms for trial prep, including LA CIV 244
  • Uploaded court documents (UD-105 Answer, complaint)
  • Possibly, Case data from LegalServer or court docket APIs
  • User-provided information

Evaluation Rubric

Signs of High-Quality Output:

  • Correct selection of all relevant CACI/specialized instructions based on defenses
  • Accurate population of case details (names, genders, pronouns, dates)
  • Clean, court-approved formatting with instructions on separate pages
  • Clear flagging of any ambiguous or missing information
  • Efficient review and finalization by staff or user
  • Trial Binder-ready output with all supporting documents included

Signs of Poor Output:

  • Missing or incorrect jury instructions
  • Misinterpretation of legal defenses
  • Formatting errors (merged instructions, missing pages)
  • Incorrect gendering, names, or case data
  • Failure to guide user through key review questions
  • Incomplete binder documents or late generation

Use Scenarios

  1. Legal Aid Workflow: Paralegal uploads client’s documents → Copilot extracts defenses and generates customized instructions → Attorney reviews → Binder printed before trial.
  2. Self-Help Litigant: Tenant receives trial date or is in the middle of finalizing the UD-105 Answer → Uploads UD-105 documents or fills in form-based intake → Copilot generates all trial docs → User downloads PDF to file and serve.