← Sticker Wall AGENT · DETAIL

EvenUp Model: AI-Generated Personal Injury Demand Letters, Cutting 7-Day Document Turnaround to 30 Minutes, Reaching a $2B Valuation

Workflow: Law firms upload raw case files such as medical records, accident reports, billing, and police reports. The vertical mod

AGENT

Key Fields

FIELD STAMPS

IndustryFintech

RegionUS（北美）

ScaleSME

ChannelOnline

🔧 Workflow

Law firms upload raw case files such as medical records, accident reports, billing, and police reports. The vertical model automatically extracts injury details, treatment trajectories, fee breakdowns, and liability highlights, generating a first draft of the demand letter, damages calculation, and medical timeline within 30 minutes. Internal legal and medical experts then perform quality reviews before delivery to the law firm, directly interfacing with insurance companies to negotiate settlement amounts. The inputs are unstructured case PDFs and files, and the outputs are claim documents and settlement strategy recommendations ready for submission to insurance companies.

🛠 Setup Requirements

Requires a background in personal injury law or partnering with attorney partners for quality control. The core is to accumulate industry document templates, settlement rule libraries, and case law corpora to train specialized terminology, cooperating with large models for structured extraction and document generation. Cold start takes about half a year, during which a small internal human review team with legal and medical backgrounds must be self-built to ensure document compliance, while continuously fine-tuning templates and prompts using real-case feedback.

🧰 Toolchain

🔧 Self-developed vertical model Piai
🔧 Large language model API
🔧 Medical record structured extraction tool
🔧 Case management system
🔧 Human legal review team

💰 Revenue

① Company level - Law firm subscription by seat/case (main revenue): 2,000 law firms × annual fee of approx. $50k per firm = ARR of approx. $100M, covering 20% of the Top 100 personal injury law firms, accounting for approx. 100% of company revenue (estimated based on financial report data; this channel's proportion was not broken down separately); ② Replicator level - Small and medium law firms pay-per-case pre-litigation as a service: Per-case fee is not public, saves an average of approx. 15 hours per case, increases settlement amounts by an average of approx. 30%, handled case volume is unvalidated, and the proportion of replicator revenue has no basis (company disclosed info); ③ Value-added/Ecosystem - AI plus human review all-inclusive delivery, moving from pure software to service delivery, delivery unit price not disclosed, and proportion likewise has no figures (publicly disclosed by company); ④ Opportunity point - Exporting settlement strategies and data to the insurance claims side, how much can be charged is not yet publicly available.

💸 Cost

Primarily model training and inference computing power, human resources for a legal and medical review team of over 100 people, and law firm customer acquisition expenses.

⏱ Time Investment

Teams operate round-the-clock, with machine processing taking about 30 minutes per case and human review taking several hours.

🚀 Getting Started

Novices should not directly copy the heavy-asset model, but can cut in with a light service providing AI-assisted medical record summaries and demand letter drafts for local small personal injury law firms. First, run a single order using general large models plus proprietary prompt templates to validate the law firm's willingness to pay before considering subscription or pay-per-case pricing. The second step is to build up your own document template library and review scripts, tie up with one or two clinics or lawyer partners for case endorsement, and then gradually expand to result-based revenue sharing.

🔑 Keys to Success

✅ Lock onto a single high-value case type to go deep, rather than building a general legal assistant
✅ AI handles the first draft, human legal experts make the final judgment to ensure documents are court-ready and compliant
✅ Use processing volume to accumulate an industry data flywheel, continuously raising document persuasiveness and settlement caps
✅ Extend from software subscriptions to all-inclusive per-case services, packaging AI efficiency and human review into a closed-loop that law firms can outsource

⚠️ 风险

⚠️ Errors in legal documents will trigger professional liability and litigation risks, requiring mandatory human review and audit trails
⚠️ Strict restrictions exist in various countries on providing legal services without a qualification, posing compliance hurdles for cross-border replication
⚠️ Large law firms building similar capabilities in-house or improvements in general model capabilities will compress premiums, requiring continuous defense through data barriers

📌 Real Cases

📌 EvenUp was founded after being rejected by YC three times, with cumulative financing exceeding $385 million. In October 2025, it completed a $150 million Series E, doubling its valuation to $2 billion, serving over 2,000 U.S. law firms, handling over 200,000 cases, and helping recover over $10 billion in compensation.
📌 EvenUp's self-developed model compresses demand letter writing from about 7 days to about 30 minutes, saving an average of 15 hours of manual work per case.
📌 As of the end of 2024, it has delivered about 100,000 demand packages and medical chronologies, helping over 1,000 law firms recover over $1.5 billion in compensation, with its vertical model reportedly far outperforming GPT-4 on multiple legal tasks.

Sources