← Sticker Wall AGENT · DETAIL

Litigation Outcome Prediction Briefing Shop: AI-Generated Win-Rate and Strategy Report Subscription, Monthly Income Approx. 25,000 RMB

Workflow: Every morning, receive case elements, causes of action, evidence lists, and jurisdiction information submitted by client

AGENT

Key Fields

FIELD STAMPS

IndustryContent / Creator Economy

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

Every morning, receive case elements, causes of action, evidence lists, and jurisdiction information submitted by clients, use legal AI tools to search public judgment document databases for similar cases, and extract judicial tendencies, compensation ranges, and key disputes. At noon, use general-purpose large language models to draft the initial report according to a fixed template, including win-rate ranges, key risk points, strategic recommendations, and settlement negotiation chip lists. In the afternoon, manually verify one by one whether the cited case numbers, courts, judgment dates, and core judgments actually exist, delete suspected hallucination content, and format it into a PDF briefing for delivery. The input consists of case facts, client question lists, and similar case databases; the output is a prediction report with citation numbers and confidence descriptions, and each report retains manual review records for accountability and review.

🛠 Setup Requirements

Requires a legal foundation or legal professional background to qualify for the review role, such as passing the bar exam, or having legal counsel or law firm assistant experience. Technically, one only needs to be proficient in using one or two legal AI search and documentation tools, mastering prompt templates and citation verification processes, with no need for self-developed models. Setup takes about one week: the first day involves selecting the primary cause of action and collecting over 100 similar case judgments as a benchmark database; the second to third days involve building report templates and verification checklists; starting the fourth day, compare AI output hallucination rates against public case number databases, solidifying the verification checklist into standard operating sheets. Overall startup costs mainly consist of tool subscriptions and personal time.

🧰 Toolchain

🔧 Legal AI search and similar case analysis tools (such as Nureal, LawGeex-like contract review and similar case tools)
🔧 General-purpose large language models (such as Claude, GPT, DeepSeek, used for drafting initial report drafts)
🔧 Public databases of judgment documents and statutory regulation databases
🔧 Document layout and PDF export tools

💰 Revenue

1. Monthly subscriptions from small and medium-sized law firms and corporate legal departments (main revenue): clients pay a monthly subscription fee, 2,000 to 5,000 RMB/client/month × serving 5 to 10 clients = monthly subscription revenue of 10,000 to 50,000 RMB, accounting for about 40%-100% of monthly revenue (estimated based on 5 clients × 2,000 to 5,000 RMB; based on case-source caliber, not independently verified); 2. Single expedited reports: law firms and legal counsel pay per request, 300 to 800 RMB per report, with public figures for monthly expedited order volume and its proportion of total revenue not specified; 3. Deep-tier value-added: stacking a deep tier containing trial strategy memos within the subscription, price included in the 2,000 to 5,000 RMB tier, not published separately, unverified client count, and its share in the total pie is also unstated; 4. Opportunity item - layered self-service subscription SaaS by cause of action: benchmarking domestic legal AI tools starting from 200-500 RMB/month (platform published pricing), turning similar case base tables into self-service query subscriptions; the scale it can achieve lacks empirical evidence.

💸 Cost

Legal AI tool subscription fees are about 500 to 1,200 RMB per month, general-purpose LLM API call fees are 200 to 600 RMB per month, access fees for judgment documents and regulation databases vary by usage from several hundred RMB per month, totaling about 800 to 2,000 RMB; working from home incurs no workspace costs.

⏱ Time Investment

About 3 hours per day, including about 1 hour for searching and AI generation, 1.5 hours for manual verification and review, and 0.5 hours for client communication; outputting 8 to 15 briefings per week, with an additional 2 to 3 hours required at the beginning of the month for renewal communication.

🚀 Getting Started

The first step for a newcomer is to pick a familiar cause of action (such as labor disputes, sales contract disputes) from public databases like China Judgment Online, organize 100 judgments to extract judgment outcomes, compensation ranges, and reasoning to build a benchmark base table; the second step is to use this base table to produce 3 free prediction samples for familiar lawyers or corporate legal counsel to try out, verifying whether the report framework is recognized; the third step is to finalize pricing and delivery templates, making a low-priced initial entry into local small and medium-sized law firms, and after collecting feedback, raise the per-customer unit price and lock in subscriptions.

🔑 Keys to Success

✅ Every citation must be manually verified for case number, court, and judgment date to prevent AI- hallucinated judgments from flowing into deliverables; this is the lifeline of the industry.
✅ Highly structured report templates (five-section format: fact summary, dispute points, win-rate range, risk list, strategic recommendations) to ensure batch replication and stable quality.
✅ Focus on a single cause of action to build expertise barriers and private similar case benchmark databases, avoiding inaccurate predictions caused by generalizing to all causes of action.
✅ Tiered client management: basic subscriptions drive volume to secure cash flow, deep strategy reports raise unit customer value, and expedited orders serve as profit supplements.

⚠️ 风险

⚠️ If outputs are used directly as legal opinions without professional review, there is a risk of misleading clients, facing complaints, or even incurring infringement liability; information assistance positioning must be explicitly stated in agreements.
⚠️ Legal AI tool hallucinations resulting in non-existent cited cases, once delivered, damage reputation; a dual-verification mechanism must be established.
⚠️ Regulatory red lines exist regarding non-lawyers providing legal opinions; cooperation with licensed lawyers or explicit clarification that reports are for internal decision-making reference only is required to prevent out-of-scope operations.

📌 Real Cases

📌 As a pioneer in contract review AI, LawGeex can already annotate risk clauses for dozens of contract types such as NDAs, SaaS agreements, and employment contracts within 30 seconds by 2026, proving that legal task-oriented AI delivery has matured (Butegou 2026 Review)
📌 Industry observations state that by 2026, about 70% of lawyers use AI in daily work and Harvey is valued at $11 billion, fully validating the willingness of legal practitioners to pay for AI tools (Sohu Legal AI Observation Article)
📌 Legora claims 2026 is the Agent year of legal AI, where intelligent agents can end-to-end autonomously complete complex legal work under human supervision, supporting individuals to undertake such outsourcing in the role of human arbiters (Legora Official Blog)
📌 Industry research shows mid-sized enterprises handle 2,000 to 5,000 contracts annually, and Agents are reshaping the full chain from drafting to archiving, leaving substantial data and review outsourcing space downstream (CSDN Contract Management AI Article)

Sources