AI Usage & Responsible AI Policy

1 Purpose of this Policy

DonePact is an AI-native agreement operations and workflow platform designed to help organizations collaborate before, during, and after execution of agreements.

The Platform may use:

• artificial intelligence
• machine learning systems
• workflow automations
• semantic indexing and retrieval systems
• generative AI models
• AI Agents
• orchestration systems
• recommendation systems
• analytics engines.

This Policy explains:

• how AI features may operate
• the role of human oversight
• limitations of AI-generated outputs
• acceptable and prohibited use of AI functionality
• responsibilities of customers and users
• DonePact’s responsible AI commitments.

2 AI Features & Capabilities

The Platform may provide AI-assisted functionality including:

• contract drafting assistance
• clause recommendations
• summarization
• metadata extraction
• semantic search
• obligation tracking assistance
• workflow recommendations
• approval routing assistance
• risk flagging
• negotiation support
• analytics
• workflow automations
• AI Agent-assisted collaboration
• operational insights
• document classification
• natural language querying.

Certain functionality may involve:

• AI-generated outputs
• automated workflows
• recommendations
• integrations with third-party systems
• customer-configured automations
• AI Agent-triggered actions.

Features may evolve over time.

3 Human Oversight Requirement

3.1 Human-in-the-Loop Principle

DonePact is designed as a human-assisted platform. AI systems and AI Agents are intended to assist human users and organizational workflows and are not intended to replace independent:

• legal judgment
• business judgment
• compliance review
• professional advice
• operational oversight.

3.2 Customer Responsibility

Customers and Authorized Users remain solely responsible for:

• reviewing AI-generated outputs
• validating accuracy and completeness
• supervising workflows
• approving contracts and decisions
• reviewing AI recommendations
• supervising AI Agent actions
• ensuring compliance with law and internal policies.

No agreement, approval, workflow decision, or legal determination should be relied upon solely on the basis of AI-generated outputs without appropriate human review.

4 AI Limitations & Disclaimers

4.1 AI Outputs May Be Imperfect

AI-generated outputs may:

• contain inaccuracies
• contain hallucinations
• omit important information
• misinterpret context
• generate incomplete analysis
• produce incorrect recommendations
• produce outdated or inconsistent responses

4.2 No Legal Advice

DonePact does not provide legal advice, legal opinions, or legal representation. AI-generated content does not constitute:

• legal advice
• regulatory advice
• professional advice
• legal opinions
• compliance certification.

Customers should consult qualified professionals where appropriate.

4.3 No Guarantee of Compliance

Use of AI-assisted features does not guarantee:

• legal compliance
• regulatory compliance
• enforceability of agreements
• risk elimination
• accuracy of outputs
• completeness of analysis.

5 AI Agents & Automated Workflows

5.1 AI Agents

The Platform may utilize AI Agents and workflow orchestration systems capable of:

• routing workflows
• generating notifications
• extracting information
• organizing workflows
• assigning tasks
• generating recommendations
• interacting with integrated systems
• facilitating operational workflows.

5.2 Customer-Controlled Automations

Customer organizations control:

• workflow configurations
• integration settings
• automation rules
• permissions
• routing logic
• AI Agent instructions
• approval structures.

Customer organizations remain solely responsible for reviewing and supervising such configurations.

5.3 Automated Actions

AI-assisted workflows and automations may:

• trigger reminders
• route approvals
• synchronize data
• update workflow statuses
• generate tasks
• send communications
• recommend actions.

Customers acknowledge that such automations may produce unintended or inaccurate results and require human oversight.

6 Third-Party AI Providers

Certain AI functionality may utilize third-party providers, including:

• AI model providers
• cloud providers
• inference providers
• semantic indexing infrastructure
• workflow infrastructure providers.

DonePact may use commercially reasonable contractual and technical safeguards designed to protect Customer Data processed through such providers.

However, Customer acknowledges that:

• AI infrastructure may involve third-party processing
• third-party systems may evolve over time
• AI models may produce variable outputs
• external AI systems may have independent operational limitations.

7 Customer Data & AI Training

7.1 Customer Ownership

Customer retains ownership of Customer Data processed through the Platform.

7.2 Restrictions on Training Public Models

Unless expressly authorized in writing by Customer:

• DonePact does not use Customer confidential content or contracts to train generalized public AI models for unrelated third parties
• DonePact does not knowingly permit third-party AI providers to independently train public models using Customer confidential content processed through the Platform.

7.3 Internal Improvement & Security

DonePact may use aggregated, anonymized, de-identified, or operational information to:

• improve platform functionality
• maintain security
• detect abuse
• improve workflow performance
• optimize operational systems
• develop features.

8 Acceptable AI Usage

Customers and users may use AI-assisted functionality solely for lawful business purposes consistent with applicable agreements and laws.

Permitted use includes:

• workflow assistance
• agreement collaboration
• operational automation
• information retrieval
• contract intelligence
• drafting support
• enterprise workflow management.

9 Prohibited AI Usage

Users shall not use the Platform or AI functionality to:

• violate applicable law
• infringe intellectual property rights
• generate unlawful, fraudulent, or deceptive content
• create malicious code or malware
• conduct phishing or impersonation
• generate harmful or abusive content
• conduct unauthorized surveillance
• process data without lawful authorization
• attempt to extract system prompts or model instructions
• reverse engineer AI systems
• benchmark or scrape AI systems for competing model training
• conduct adversarial attacks or prompt injection attacks
• bypass security or access controls
• use the Services to train competing AI systems without authorization.

DonePact may suspend or restrict access for violations of this Policy.

10 Data Security & Confidentiality

DonePact implements commercially reasonable technical and organizational safeguards designed to protect Customer Data and AI processing activities.

Security measures may include:

• encryption
• access controls
• authentication measures
• logging and monitoring
• tenant isolation
• internal confidentiality obligations
• infrastructure security controls.

Customers remain responsible for:

• managing user permissions
• supervising integrations
• reviewing workflow configurations
• securing credentials
• supervising AI-assisted actions.

11 Beta, Experimental & Early Access Features

Certain AI features may be designated as:

• beta
• preview
• experimental
• early access
• pilot.

Such features may:

• change over time
• contain bugs or inaccuracies
• be incomplete
• be interrupted
• be discontinued
• produce inconsistent outputs.

Beta or experimental AI features are provided “AS IS” without warranties of any kind. Customers use such features at their own risk.

12 Responsible AI Principles

12.1 Human-Centered Design

AI systems should support human decision-making rather than replace responsible human oversight.

12.2 Transparency

Users should understand that AI-generated outputs may be probabilistic and imperfect.