# Discovery Review Agent

*/Opportunities/Discovery_Review_Agent*

## Opportunity Overview

**Wedge**: The beachhead targets privilege log generation for medium-stakes commercial litigation. This niche relies on highly defined, rule-based standards that are tedious for humans but easily codifiable for an agent, offering immediate cost savings and risk reduction. After proving reliability in identifying and logging privileged communications, the product expands into complex responsiveness coding and ultimately synthesizes key documents for deposition preparation.
**Timing**: Large context window LLMs capable of handling over one million tokens now allow agents to retain complex case narratives and apply them across gigabytes of unstructured emails and PDFs, achieving accurate document classification without the severe context-loss seen in earlier models.
**Why This I C P**: Mid-size litigation boutiques handle data-heavy commercial disputes but lack the associate headcount of BigLaw, giving them a strong financial incentive to adopt technology that replaces outsourced review vendors and increases their per-partner leverage.
**Size Of Prize**: Approximately 10,000 mid-to-large litigation firms and corporate legal departments in the US spend an average of $100,000 annually on contract attorney labor for first-pass document review, yielding an addressable market of roughly $1B.
**Gap Narrative**: First-pass discovery review demands thousands of hours of contract attorney labor to code documents for responsiveness and privilege. Current eDiscovery platforms rely on keyword search and predictive coding but still require human reviewers to read borderline materials and apply case-specific logic. Litigation practices need an autonomous system that ingests a case strategy memo and directly outputs coded document sets with cited reasoning.
**Defensibility**: Defensibility compounds through workflow lock-in and model personalization. As the agent observes corrections from senior attorneys, it learns the specific relevance thresholds and subjective coding preferences of the firm, creating high switching costs since a competing vendor would require starting the training process over from scratch.
**Why This Thesis**: The Agent thesis matches this problem because first-pass review is an autonomous execution task, not a software workflow to be managed. Firms want to delegate the classification work entirely by providing instructions and receiving completed work product, which an autonomous agent delivers directly.

## Opportunity Linked Thesis

**Thesis**: [Agent](/Theses/Agent)

## Opportunity Linked I C P

**Icp**: [Litigation Law Firm](/CompanyTypes/Litigation_Law_Firm)

## Opportunity Market Sizing

_Illustrative — target and order-of-magnitude estimate figures, not an achieved track record (this Thing is concept-stage)._

**S A M**: ~$500M-800M US mid-market litigation boutiques and regional law firms
**S O M**: ~$15M-40M
**T A M**: ~50,000 US litigation law firms and corporate legal departments × ~$40,000/yr average agent spend ≈ ~$2B
**Growth Rate**: ~15-20%/yr, driven by exploding enterprise data volumes across internal collaboration applications increasing the average discovery corpus size
**Paid Comparable Spend**: ~$150k-500k+/yr per firm on outsourced contract attorneys for managed review and per-GB legacy e-discovery platform hosting fees

## Opportunity Incumbents

- [RelativityOne Platform](/Products/RelativityOne_Platform) — Tool
- [Everlaw E-Discovery](/Products/Everlaw_E-Discovery) — Tool
- [CS Disco Platform](/Products/CS_Disco_Platform) — Tool
- [Managed Review Providers](/Products/Managed_Review_Providers) — Service
- [Manual Paralegal Review](/Products/Manual_Paralegal_Review) — DIY
- [Excel Review Logs](/Products/Excel_Review_Logs) — Spreadsheet

## Opportunity Win Conditions

**Kill Thresholds**:
- Human override rate > 12% on privilege tags after 45 days
- Inference and processing cost > $0.15 per document
- Zero deployments on active litigation datasets > 50GB within 90 days
- D60 retention < 30% across newly onboarded litigation boutiques
**Leading Metrics**:
- Human override rate on privilege and relevance tags
- Time from raw data ingestion to first classified document batch
- Inference cost per 1,000 documents classified
- Percentage of firm matters utilizing the agent over manual review
**What Proves Right**: Litigation teams deploy the agent on active discovery corpora exceeding 100GB and accept the system's relevance classifications without full manual re-review. Firms successfully bill the agent's processing time to clients as a flat fee or technology disbursement, securing a clear margin over inference costs. Cohorts retain usage across consecutive case matters, proving the platform directly replaces outsourced contract attorney spend.
**What Proves Wrong**: Partners refuse to trust the agent's privilege classifications, forcing paralegals to manually verify every document and eliminating the margin advantage. Per-document inference costs escalate beyond the equivalent hourly rate of outsourced contract reviewers, destroying the economic incentive. Integration barriers with legacy e-discovery platforms prevent teams from exporting their raw data corpus into the agent.

## Opportunity Build Profile

**Hardest Part**: Maintaining >95% recall on relevance and near-perfect precision on privilege identification across massive, unstructured document dumps without dropping context or hallucinating legal rationale.
**Min Viable Scope**: Build a first-pass privilege identifier strictly for English-language email chains (.pst files) and standard PDFs. Deliberately exclude multimedia transcription, optical character recognition for handwritten notes, and final production formatting.
**Cold Start Problem**: Firms block access to raw discovery data due to strict confidentiality rules and inherent distrust of new review tools. Break this by benchmarking the system against standard public datasets like the Enron emails, then offering an air-gapped shadow run parallel to associates on a closed case.
**Time To First Value**: 24 hours to ingest a raw document dump and generate a verifiable, first-pass privilege log
**Data Moat Available**: true
**Technical Difficulty**: High

## Neighborhood

### Where the gap lives

- [Boutique Law Firm  NAICS-5411](/CompanyTypes/Boutique_Law_Firm__NAICS-5411) — latent gap · CompanyTypes

### Incumbent in

- [RelativityOne Cloud Platform](/Products/RelativityOne_Cloud_Platform) — incumbent in · Products
- [Everlaw Discovery](/Products/Everlaw_Discovery) — incumbent in · Products
- [CS Disco](/Products/CS_Disco) — incumbent in · Products
- [Excel Review Logs](/Products/Excel_Review_Logs) — incumbent in · Products
- [Managed Review Providers](/Products/Managed_Review_Providers) — incumbent in · Products
- [Manual Paralegal Review](/Products/Manual_Paralegal_Review) — incumbent in · Products

### Applies thesis

- [Litigation Law Firm](/CompanyTypes/Litigation_Law_Firm) — applies thesis · CompanyTypes

### Embodies

- [Agent](/Theses/Agent) — embodies · Theses

### Similar Opportunities

- [First-Pass Discovery Agent](/Occupations/Legal_Occupations/Opportunities/First-Pass_Discovery_Agent) — similar · Opportunities
- [Discovery Triage Agent](/Opportunities/Discovery_Triage_Agent) — similar · Opportunities
- [Blue-Chip Discovery Agent](/Opportunities/Blue-Chip_Discovery_Agent) — similar · Opportunities
- [Iterative Discovery For Counsel](/Opportunities/Iterative_Discovery_For_Counsel) — similar · Opportunities
- [E-Discovery as a Service](/Opportunities/E-Discovery_as_a_Service) — similar · Opportunities
- [E-Discovery Review Engine](/Opportunities/E-Discovery_Review_Engine) — similar · Opportunities
- [E-Discovery as a Service](/Knowledge/Law_and_Government/Opportunities/E-Discovery_as_a_Service) — similar · Opportunities
- [Autonomous Video Discovery](/Opportunities/Autonomous_Video_Discovery) — similar · Opportunities
- [Contract Redlining Agent](/Opportunities/Contract_Redlining_Agent) — similar · Opportunities
- [E-Filing Attribution Agent](/Opportunities/E-Filing_Attribution_Agent) — similar · Opportunities
- [Automated Contract Triage](/Opportunities/Automated_Contract_Triage) — similar · Opportunities
- [Case Data Router](/Opportunities/Case_Data_Router) — similar · Opportunities
- [Brief Validation Agent](/Opportunities/Brief_Validation_Agent) — similar · Opportunities
- [Rate Case Agent](/Opportunities/Rate_Case_Agent) — similar · Opportunities
- [Legal Intake Agent](/Opportunities/Legal_Intake_Agent) — similar · Opportunities
- [Retention Compliance Agent](/Opportunities/Retention_Compliance_Agent) — similar · Opportunities
- [Scan Router](/Opportunities/Scan_Router) — similar · Opportunities
- [Intake Triage Agent](/Opportunities/Intake_Triage_Agent) — similar · Opportunities
- [Process Server Dispatcher](/Opportunities/Process_Server_Dispatcher) — similar · Opportunities
- [Litigation Underwriting Engine](/Occupations/Legal_Occupations/Opportunities/Litigation_Underwriting_Engine) — similar · Opportunities