← Sticker Wall AGENT · DETAIL

Law Firm Batch Judgment Dispute Focus Mind Map & Similar Case Comparison Matrix Pay-per-Article Model Earning 5,000 RMB Monthly

Workflow: Lawyers or legal assistants send a batch of judgments with the same cause of action in HTML, PDF, or Word formats along

AGENT

Key Fields

FIELD STAMPS

IndustryAI / LLM

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

Lawyers or legal assistants send a batch of judgments with the same cause of action in HTML, PDF, or Word formats along with retrieval requirement tags via WeChat, email, or Feishu forms. The AI automatically calls large model APIs to complete structured parsing of the judgments, extracting dispute focuses, court viewpoints, legal reasoning, and judgment results. It generates a Dispute Focus Mind Map Markdown file and a Similar/Different Case Comparison Excel matrix, finally outputting a reviewable Word report for manual final review and signature. Handle one to two batches of customer deliverables daily, with 5 to 10 articles per batch, priced at 300 to 500 RMB per article, delivered after customer confirmation.

🛠 Setup Requirements

Requires registering Claude or Tongyi Qianwen LLM API keys to obtain structured extraction capabilities; using Python or n8n to build the judgment HTML parsing and legal entity recognition pipeline; preparing python-docx and xlsxwriter libraries for generating Word reports and Excel comparison matrices; mastering basic legal terminology and judgment structure knowledge, understanding standard judgment formats; referencing the legal-case-report project by krionwu23 on GitHub as a technical baseline, reusing its HTML parsing and structured extraction concepts; overall setup cycle takes about one to two weeks, with the first full workflow run taking 3 to 5 days to debug prompts and parsing rules.

🧰 Toolchain

🔧 Claude API or Tongyi Qianwen API
🔧 n8n low-code workflow or Python scripts
🔧 python-docx and xlsxwriter office document generation libraries
🔧 GitHub legal-case-report project as technical reference

💰 Revenue

① Pay-per-article by small and medium law firm legal assistants, intern lawyers, and corporate legal counsel (main revenue): clients pay per article, 300-500 RMB/article × 10-15 articles processed stably per month = monthly income of about 3,000-7,500 RMB, disclosing about 5,000 RMB monthly, contributing 100% of the revenue (estimated, based on merchant statements, independent review pending); ② Complex major cases and batch commissions per project (project service fees): charged at 800-1,500 RMB/project × number of undertaken projects unverified, project service proportion unspecified (merchant statement, third-party review lacking); ③ Monthly subscription for structured case retrieval (subscription model): providing law firms with structured statute and case retrieval services via monthly subscription, subscription price undisclosed, number of contracted law firms unverified, case retrieval contribution proportion unclear; ④ Opportunity item: batch judgment structured annotation via volume API, the proportion of this batch annotation path remains suspended.

💸 Cost

LLM API call costs are about 300 to 500 RMB per month, billed based on the number of processed judgment pages; Claude Pro subscription at $20 per month for deep analysis of complex cases; domain name and cloud function deployment costs of 50 RMB per month; total fixed costs are about 600 to 800 RMB per month, with marginal costs close to zero and gross margins exceeding 90%.

⏱ Time Investment

Investing 1 to 2 hours daily, mainly handling client deliverables, manual final review of AI-generated results, and formatting adjustments; backlog cases can be batched over weekends; time is flexible and controllable, making it suitable as a side hustle or a freelancer's primary business without taking up too much daytime working hours.

🚀 Getting Started

Step 1: Search and clone krionwu23's legal-case-report project on GitHub to learn its complete workflow from judgment HTML to structured Word reports, understanding HTML parsing rules and legal entity extraction logic. Step 2: Set up a local Python environment, configure the n8n workflow, and test with 3 to 5 public judgments from China Judgments Online to verify dispute focus extraction accuracy and similar/different case identification capabilities. Step 3: Refine and package the similar case comparison matrix template, find 1 to 2 lawyer friends or legal interns for trial feedback, continuously optimize prompts and parsing rules, and after maturing, take orders in Xiaohongshu legal communities, lawyer platforms, or legal tech communities.

🔑 Keys to Success

✅ Judgment HTML parsing accuracy and traceability of original text citations to build client trust
✅ Dispute focus extraction capability must distinguish between case facts and legal application to avoid AI hallucinations and misinterpretation of judicial intent
✅ Excel format of the similar and different case comparison matrices must facilitate direct citation by lawyers in proxy statements, defense briefs, or case analysis reports
✅ Maintain key manual final review and signature steps to ensure the professionalism and compliance of legal documents and mitigate practice risks
✅ Build lawyer client word-of-mouth and repurchase mechanisms, enhancing client retention through pay-per-article subcontracting models or monthly subscription systems

⚠️ 风险

⚠️ Judgment HTML formats vary, with significant differences in web structures across courts, requiring continuous adaptation of parsing rules leading to increased maintenance costs and delivery delays
⚠️ Dispute focus extraction involves professional legal judgment; AI may misinterpret judicial intent or omit key reasoning, requiring strict manual review
⚠️ Lawyer clients have varying levels of acceptance regarding the credibility of AI-generated content, and some senior lawyers may refuse to use AI-assisted reports
⚠️ The legal industry has high compliance requirements; attention must be paid to data confidentiality and client privacy protection to avoid practice risks

📌 Real Cases

📌 Xiangyu Workflow Legal Knowledge Base AI Skill: Contains an 8.5-million-word statute and case library, providing structured statute and case retrieval references; its knowledge base construction concept and structured output format can serve as a template reference for this project
📌 Wulv AI Platform: Achieves a 30-minute completion of online case filing services that originally took 15 days across the full process, verifying market demand and payment feasibility for legal AI tools based on search results
📌 AI Legal Assistant Side Hustle Case: Claude contract review side hustle reported by DuckDB Lab earning 5,000+ RMB monthly, proving that the monetization path for legal AI tools is feasible and lawyer user payment willingness is clear
📌 GitHub Open Source Project legal-case-report: Generates reviewable Word reports and audit results from judgment HTML, with contributions from open-source community members, proving the technical implementation path is feasible

Sources