Modern Review Starts Here
Simple Integration
No steep learning curves or complex integrations. Forest is a flexible, intuitive GenAI application that empowers legal teams to streamline review with confidence, control, and clarity.
GenAI Meets Human Oversight
Forest is a tech-forward approach built for legal workflows, combining prevalent large language models (LLMs) with a human-centered design. With direct and dynamic user interaction, legal teams have complete control over AI-assisted outputs without sacrificing precision or oversight.
Expert Guided
Our in-house AI & Analytics consultants deploy Forest on a client’s behalf, ensuring client AI workflows are effective and defensible. With expert-tuned prompts and human-in-the-loop support, clients also enjoy significant cost savings without sacrificing accuracy.
Learn More About Forest Ai
Intuitive. Dynamic. Efficient.
Forest is a comprehensive application integrated into the Relativity review platform, using the capabilities of OpenAI’s GPT-4o and GPT-4o mini LLMs for multiple use cases.
Translation & Interpretation
Progress beyond literal translation with AI-assisted interpretation that captures context, nuance, and cultural idioms. Forest’s customizable prompts enable teams to surface multiple interpretations for ambiguous language or idiomatic phrases, revealing intent and subtext that traditional tools or even native speakers may overlook. Results can be validated or refined by human linguists to ensure clarity, accuracy, and legal defensibility.
- Ideal for multilingual litigation, cross-border matters, regulatory and nuanced internal communications
- Capture tone, intent, and multiple meanings
- Customizable prompts for idiom and ambiguity resolution
Privilege Log Drafting
Accelerate privilege log drafting through structured, AI-assisted summaries aligned with your existing protocols. Forest is more economical than aiR, with potential savings of up to 40%, even when undertaking a 100% human review of the Forest-drafted log lines. With additional built-in efficiency, Forest supports stringent sampling workflows to reduce touchpoints while maintaining accuracy.
- The best of tech and humans
- Built for teams that demand precision
- Transparent, auditable, and DOJ-ready
Document Summarization
Transform long-form documents into clear, actionable summaries tailored to your guidance. Forest surfaces key facts, identifies tone or sentiment, and supports cross-border matters with concise and accurate insights – even across multiple languages. Key documents can also be automatically slotted into timeline views, giving reviewers and attorneys instant chronological context.
- Cross-language summarization
- Efficient and reliable context building
- Timeline-ready outputs for critical documents
Relevancy Determination
Classify documents based on case-specific background, key players, and relevancy issues or RFPs. Forest distinguishes truly responsive content from non-responsive or marginal material by using tailored prompts, and sentiment and reasoning analysis. Edge cases and low-confidence predictions are automatically flagged for human review to establish defensibility even in nuanced or high-stakes matters.
- Case-tuned classification workflows
- Transparent reasoning and sentiment detection
- Triage for focused or time-sensitive reviews
PII & Sensitive Information Detection
Identify multiple customizable categories of PII, business-sensitive data, and regulatory triggers (including export-controlled materials) directly from unstructured text. With a fully customizable process, legal teams can respond quickly to a breach response or compliance audits.
- Support for breach review, regulatory reporting and redaction identification
- High-precision identification in structured and unstructured text
- Flexible extraction formats for some data types for downstream workflows
Case Studies
Challenge: A national law firm was preparing a privilege log for a fast-moving investigation with over 5,000 privileged documents. Traditional manual logging would have required significant attorney time and risked challenge to the formulaic descriptions.
Solution: Using Forest, the review team entered their privilege review protocol, designated attorney names and roles, and specified the output format required by the court. Forest then generated modular, fielded log entries including:
- Document type
- Privileged action and description
- Legal entities involved
- Subject matter (“regarding”)
These components were automatically structured into separate fields, enabling mass editing and batch updates. When one of the listed attorneys changed her last name midreview the team updated a single field — and Forest automatically applied the change across all relevant entries, eliminating hundreds of manual edits.
Results:
- 70% reduction in time-to-log versus manual drafting
- Consistent, diverse, and defensible descriptions across all entries
- Streamlined quality control and rapid responsiveness to necessary changes
Challenge: A litigation team was facing a tight deadline: 1,700 key documents had been identified for review ahead of a high-stakes deposition cycle. Senior attorneys needed to understand each document’s significance — and how they fit together chronologically — without reading every page.
Solution: Using Forest, the team generated structured summaries for each document, guided by matter-specific instructions and issue framing. The output included:
- A concise, three-sentence summary of each document
- Highlighted key facts and quoted language
Forest then populated a timeline view using these summaries, allowing attorneys to quickly explore the document set by date, theme, or custodial source. No embeddings or training were required — just consistent prompt logic.
Results:
- 1,700 documents summarized in less than a day
- Chronological and thematic context delivered in an intuitive, web-based timeline with links back to the source documents
- Enabled senior attorneys to spot gaps, contradictions, and key admissions before depositions began
Challenge: In high-volume matters where manual responsiveness review is time- and cost-prohibitive, Forest enables a prompt-driven, AI-assisted relevance workflow modeled on traditional TAR 1 principles — but without the need for training sets or model tuning.
Solution: Forest was deployed to classify documents for responsiveness based on case background, custodian information, and specific RFPs or issues. The workflow included:
1. Prompt Iteration
Forest’s template prompt was refined through iterative testing and feedback from senior reviewers, incorporating reasoning, issue context, and language tone calibration.
2. Control Set Validation
A randomly selected control set of human-coded documents was held back and used to validate Forest’s output, measuring agreement rates, identifying edge cases, and assessing success.
3. Edge Case Triage
Documents flagged as borderline or low-confidence were automatically routed for human review, preserving cost savings while mitigating risk.
Results:
- Delivered TAR-like defensibility without the overhead of continuous training
- Reduced review volume by 60–80% while maintaining alignment with senior reviewers
- Created a clear audit trail of prompt iterations and validation results to support production decisions