Document Automation for Legal Aid: A Working Pattern

Legal aid clinics serving low-income clients face high document volume. Document automation helps, but only when applied thoughtfully.

Goals

1. Generate court-filed documents from client intake
2. Fill forms with client information
3. Maintain version control over templates
4. Verify output against current court rules
5. Track workflow state across cases
6. Preserve confidentiality

Stack

• docassemble (https://docassemble.org) for guided interviews
• Form templates stored as YAML or markdown
• PostgreSQL for case and client data
• Self-hosted or BAA-protected LLM endpoint
• Document assembly via ReportLab or WeasyPrint
• Audit log + encryption at rest

Implementation pattern

Intake flow

Client calls or visits legal aid intake. Intake worker enters:

• Client name and contact
• Case type (eviction, family, consumer)
• Key dates (hearing dates, deadline dates)
• Income for eligibility verification
• Brief case narrative

Eligibility check

• Income vs. federal poverty line
• Residence / citizenship status
• Case type vs. clinic scope

If eligible, proceed. If not, refer.

Document generation

• Generate applicable forms
• Fill client-specific information
• Include supporting documents
• Add procedural instructions

Lawyer review

A supervising lawyer reviews:

• Document completeness
• Procedural correctness
• Risk flags

If approved, send to client for signature and filing.

Use cases

• Eviction defense
• Family law
• Consumer protection
• Public benefits
• Employment

Eval framework

• Documents generated per month
• First-pass approval rate by lawyer
• Revision required (>1)
• Time saved per document
• Filing rejection rate

Targets:

• 90% first-pass approval
• 0% filing rejections due to automation errors
• 30 minutes saved per document

Open-source references

• docassemble: MIT
• Suffolk LITLab HoudiniCollection: MIT
• Python-docx-template: MIT
• WeasyPrint: BSD-3-Clause

Limitations

1. DocAuto does not replace attorney judgment
2. Form version drift requires constant maintenance
3. Jurisdiction-specific variation is expensive to scale

Acknowledgments

This article summarizes observed patterns from legal aid clinics. Specific deployments vary by jurisdiction.

Dillon Deutsch has built document automation systems for legal services. https://courtgpt.ai