Playbook Configuration

What is the Playbook?

The playbook is your organization's institutional knowledge base for contract review. It encodes your firm's standard positions, preferred language, risk thresholds, regulatory requirements, and terminology definitions. During document analysis, the pipeline automatically retrieves relevant playbook entries and injects them into AI prompts, ensuring that every contract is reviewed against your organization's specific standards — not generic legal principles.

Why the playbook matters

Without a playbook, Contract Lucidity performs general-purpose contract analysis. With a well-populated playbook, it becomes a system that reviews contracts the way your senior attorneys would — flagging deviations from your firm's positions, applying your risk tolerances, and referencing your preferred clause language.

The Five Categories

The playbook is organized into five categories, each serving a distinct purpose in the analysis pipeline.

1. Terminology

Purpose: Define key legal terms as your organization interprets them.

Content structure:

Field	Description
term	The legal term being defined
definition	Your organization's standard definition
usage_notes	Guidance on negotiation, carve-outs, common counterparty positions

Example: Define "Material Adverse Change" with your standard carve-outs (general economic conditions, industry-wide changes, changes in law) and note that buyers push for broad MAC definitions while sellers negotiate narrow carve-outs.

Pipeline usage: Injected during structured data extraction and clause analysis. When the AI encounters a defined term in a contract, it compares the contract's definition against yours and flags deviations.

2. Position Standards

Purpose: Document your organization's negotiating positions for specific clause types.

Content structure:

Field	Description
clause_type	The type of clause (e.g., "Limitation of Liability")
our_position	Your firm's ideal/standard position
acceptable_range	The range of terms you would accept in negotiation
walk_away	Terms that are unacceptable and would prevent signing

Example: For limitation of liability in technology agreements — standard is 12 months of fees, acceptable range is 6-24 months, walk-away is uncapped liability or a cap below 3 months.

Pipeline usage: Injected during clause analysis (all 5 categories are retrieved). The AI evaluates each extracted clause against your position standards and reports where the contract falls relative to your acceptable range.

3. Risk Thresholds

Purpose: Define what constitutes low, medium, and high risk for specific clause types.

Content structure:

Field	Description
clause_type	The type of clause being scored
green	Conditions that represent low/acceptable risk
yellow	Conditions that represent medium/elevated risk
red	Conditions that represent high/unacceptable risk
override_default	Whether this threshold overrides the AI's default risk assessment

Example: For indemnification — green is mutual indemnification subject to a cap, yellow is broad indemnification not subject to a cap, red is indemnification for consequential damages that survives indefinitely.

Pipeline usage: Injected during clause analysis and report generation. Risk thresholds directly influence the severity ratings (green/yellow/red) in the contract review report.

4. Clause Templates

Purpose: Store your preferred contract language for common clause types.

Content structure:

Field	Description
clause_type	The type of clause
template_text	Your organization's preferred language
usage	When and how to use this template, with industry-specific adjustments

Example: A force majeure clause with your preferred trigger list, mitigation obligations, and a 90-day termination right, with notes on adjusting the trigger list for financial services vs. manufacturing.

Pipeline usage: Injected during clause analysis. The AI compares contract language against your templates and can suggest your preferred language as a replacement in clause revision workflows.

5. Regulatory Context

Purpose: Encode relevant regulatory requirements that contracts must address.

Content structure:

Field	Description
regulation	Name of the regulation or statute
applicability	When this regulation applies (thresholds, jurisdictions, industries)
requirements	What the regulation requires in contractual terms
consequences	Penalties, damages, or risks of non-compliance

Example: CCPA/CPRA compliance requirements for service provider agreements, including the five mandatory contract provisions, applicability thresholds, and statutory damages.

Pipeline usage: Injected during clause analysis when the document's jurisdiction or subject matter matches. The AI checks whether required regulatory provisions are present and adequate.

How Playbook Context is Retrieved

The pipeline does not inject every playbook entry into every analysis. Instead, it uses semantic search to find the most relevant entries.

Each pipeline stage retrieves different categories:

Pipeline Stage	Categories Retrieved	Max Entries
Structured data extraction	Terminology, Position Standards	5
Clause analysis	All 5 categories	8
Report generation	Risk Thresholds, Position Standards, Regulatory Context	5

tip

Playbook embeddings are generated automatically when the pipeline reaches the playbook stage. If you add new playbook entries, they will be embedded and available for the next document processed.

Populating the Playbook

Via the UI

1. Navigate to Settings > Playbook in the sidebar
2. Select a category tab (Terminology, Position Standards, Risk Thresholds, Clause Templates, Regulatory Context)
3. Click Add Entry
4. Fill in the title, description, and category-specific fields
5. Assign tags (practice area, contract type, jurisdiction) to help with filtering and organization
6. Save the entry

Seed Data

On first boot, Contract Lucidity seeds sample playbook entries across all five categories to demonstrate the system. These include:

• Terminology: Material Adverse Change, Confidential Information
• Position Standards: Limitation of Liability, Non-Compete Duration
• Risk Thresholds: Indemnification Scope, Termination for Convenience
• Clause Templates: Force Majeure, Data Processing Addendum (Purpose Limitation)
• Regulatory Context: CCPA/CPRA Compliance, Delaware Fiduciary Duties

The seed data also creates tags for 12 practice areas, 14 contract types, and 12 jurisdictions.

warning

The sample entries are illustrative. Review and replace them with your organization's actual standards before processing real contracts.

Tagging System

Every playbook entry can be tagged across three dimensions:

Tag Type	Purpose	Examples
practice_area	Legal practice area	Corporate/M&A, Employment & Labor, IP, Real Estate
contract_type	Type of contract	NDA, MSA, SOW, Employment Agreement, Lease
jurisdiction	Legal jurisdiction	US Federal, US-NY, US-CA, UK, EU, Singapore

Tags serve two purposes:

1. Organization: Filter and browse entries in the UI
2. Context: Help the semantic search find relevant entries — entries tagged with a matching contract type score higher in retrieval

Best Practices

Terminology

• Define every term where your interpretation differs from common usage
• Include negotiation context in usage_notes — what counterparties typically push for and where to push back
• Keep definitions precise and actionable

Position Standards

• Cover your most frequently negotiated clause types first
• The walk_away field is critical — it tells the AI what constitutes a deal-breaker
• Update positions as firm policy evolves
• Be specific about numbers (e.g., "12 months of fees" not "a reasonable cap")

Risk Thresholds

• Ensure green/yellow/red criteria are mutually exclusive and collectively exhaustive
• Set override_default to true for clause types where your risk tolerance differs significantly from general legal norms
• Test thresholds against real contracts to calibrate sensitivity

Clause Templates

• Store your actual preferred language, not generic boilerplate
• Include usage notes with industry-specific variations
• Keep templates current with regulatory changes

Regulatory Context

• Focus on regulations that directly affect contract terms (GDPR, CCPA, OFAC, etc.)
• Include clear applicability criteria so the AI knows when to surface them
• Update consequences with current penalty amounts

Playbook and the Version-Aware Pipeline

When a revised version of a document is uploaded, the playbook interacts with version context:

1. Classification stage: Playbook context is injected alongside the prior version's classification, so the AI can detect if a reclassification is needed
2. Structured extraction: Playbook terminology and position standards inform which fields to re-extract when the diff touches them
3. Clause analysis: The AI receives both playbook context and attorney decisions from the prior version, enabling it to check whether revisions align with your organization's standards
4. Report generation: Risk thresholds from the playbook are applied to the new version, with explicit comparison to the prior version's risk profile

This means your playbook standards are consistently applied across all versions of a document, and the AI can detect when a revision moves a clause from within your acceptable range to outside it (or vice versa).