GrabDocs AI Policy

1. Purpose

The purpose of this policy is to:

• Ensure transparent and responsible use of AI across GrabDocs services.
• Inform users about AI functionality in document processing, chat, question-answering, classification, and analytics.
• Define rules for data handling, security, and user control when AI is applied.
• Support compliance with applicable privacy, security, and other relevant requirements.

2. Scope

This policy applies to GrabDocs services that involve AI, including but not limited to:

• AI-powered document search and retrieval (semantic and keyword search over your documents using vector embeddings and indexes).
• AI-assisted chat and Q&A across documents (smart chat with intent understanding, multi-step reasoning, and source references).
• Automated document classification, tagging, and summarization (e.g., document type classification, document profiles with keywords and categories, meeting insights and trend summaries).
• AI-based extraction of data (e.g., structured fields from forms, receipts, invoices; tables and metadata; custom extraction prompts).
• Query understanding and analytics (interpretation of natural-language questions, analytics-style queries, and metadata-style queries).
• Embedding and indexing of document content for fast, relevant search within your workspace.

3. AI Use in GrabDocs

3.1. Document Processing

• AI may extract text and metadata from user-uploaded documents for indexing, search, and analytics.
• Document content may be processed in real time or during ingestion by AI models to provide:
  • Classification (e.g., Form, Receipt, Document, Picture) and confidence scores.
  • Document profiles (e.g., title, author, summary, keywords, entities, categories) used for filtering and discovery.
  • Structured extraction where applicable (e.g., form fields, receipt line items).
• AI processing is scoped by user and workspace.
• Indexes and embeddings are tied to your workspace, so access remains subject to GrabDocs account, company, and workspace permissions.
• Where feasible, fallback logic and local rules are used first (e.g., intent classification, text-heavy document detection) to reduce unnecessary AI calls and improve consistency.

3.2. Chat and Question-Answering

• AI can answer user queries across documents and related context (such as bookmarks or transcripts) that the user is authorized to access within their workspace(s). Where applicable, local or rule-based logic may be used first for simple or metadata-style queries before calling AI.
• AI responses are generated using document content retrieved from GrabDocs' search indexes and third-party language models (see Section 5). Responses may include source references such as relevant file names or citations when available.
• Conversation context (e.g., recent messages) may be sent to a model within a request to improve follow-up answers.
• Chat history and related interaction data may be stored in GrabDocs to support continuity, history, security, support, and audit functions, as described in our privacy notice and agreements.
• Usage and rate limits may apply to AI chat and related features to ensure fair use and system stability.

3.3. Automated Classification and Tagging

• Documents may be automatically labeled or categorized by AI (e.g., document kind, categories, keywords in document profiles). Fallback logic and local rules may be used first where applicable.
• Users can override or correct AI-derived labels where the product allows. Human review is recommended for critical decisions.
• AI classifications are probabilistic and may not be fully accurate; they are intended to assist, not replace, human judgment.
• Sensitive content handling: If an AI provider declines to process certain content under its service rules, GrabDocs may return a neutral result or error rather than exposing raw provider output.

3.4. Embeddings and Search

• To power semantic search and retrieval-augmented chat, GrabDocs generates vector embeddings of document text and, where applicable, related content using configured embedding models.
• These embeddings are stored in GrabDocs' search infrastructure and are tied to your account and workspace context.
• Indexes and embeddings for a document are removed, updated, or excluded from retrieval when the document or associated data is deleted or moved out of active use, consistent with our retention and deletion practices.

4. Data Handling and Privacy

• Ownership: Users and institutions retain ownership of uploaded content and associated metadata. GrabDocs does not claim ownership of your documents or outputs derived from your data.
• Processing: AI may access document content, metadata, or relevant chunks to answer queries, generate summaries, classify content, or perform extraction. GrabDocs minimizes and scopes data sent to AI providers to the portions needed to fulfill the user's request, such as relevant document chunks rather than full documents where feasible.
• Storage: GrabDocs stores document, profile, search, and chat-related data in its systems as needed to operate the product and provide continuity, support, security, and audit features. Vector embeddings and indexes are stored within GrabDocs-managed infrastructure used for retrieval. Storage of such data is solely for operating, maintaining, securing, supporting, and providing the service, and does not transfer ownership to GrabDocs.
• Security: AI-related processing follows GrabDocs security controls, including encryption in transit and at rest where applicable, access controls, and monitoring consistent with our security practices.
• Data location: Processing and storage occur in infrastructure and regions described in our privacy notice, contracts, or deployment documentation, as applicable.
• Deletion: When you or an administrator deletes document or account data, GrabDocs removes, deactivates, or schedules deletion of corresponding data from applicable systems, including search indexes and vector embeddings, consistent with our retention and deletion practices.
• Retention: Data is retained only as long as reasonably necessary to provide the service, maintain security and support functions, and meet applicable legal or operational obligations.

5. AI Model Transparency

• GrabDocs currently uses OpenAI to provide certain AI-powered features, such as chat, summarization, classification, extraction, and semantic search. Over time, GrabDocs may also use additional AI models or service providers to support these or related features.
• GrabDocs uses third-party AI/LLM services, including OpenAI, for:
  • Language model APIs (e.g., chat, summarization, classification, extraction, and query understanding).
  • Embedding APIs (e.g., semantic search and retrieval).
• When using third-party AI services, GrabDocs sends the content needed to perform the requested functionality. How that data is handled by the provider depends on the provider's applicable service terms and arrangements with GrabDocs.
• GrabDocs does not use customer data to train foundation models and does not intentionally provide customer data to third-party AI providers for training their general models.
• AI processing respects existing user permissions and access controls. Content a user is not authorized to access is not intended to be included in that user's AI results or responses.
• AI models provide suggestions, insights, and classifications; the final decision rests with the user or institution. GrabDocs does not represent AI output as a substitute for required human judgment.
• AI-generated outputs may occasionally be inaccurate or incomplete. Users should verify important information independently and use AI as an assistive tool, not as a substitute for professional advice.

6. User Control

• Configuring AI features: Where the product supports it, certain AI-powered features may be configurable, limited, or enabled only for specific users, plans, or workspaces.
• Deletion of AI-processed data: You may request deletion of your data, including data that has been processed by AI (e.g., document profiles, indexes, or chat history), subject to our normal retention and product practices.
• Account and privacy settings: Choices related to data use, retention, and AI may be available through product settings, administrative controls, or other account-level arrangements, depending on your deployment and plan.
• Usage limits: Some plans may have token or usage limits for AI features. When limits apply, GrabDocs may enforce them before calling AI services.

7. Limitations and Disclaimers

• AI outputs are assistive tools, not legal, financial, medical, or other professional advice.
• AI may misinterpret or misclassify complex or ambiguous documents; human review is recommended for high-stakes or regulated use cases.
• Users are responsible for reviewing and verifying AI outputs before relying on them for business, legal, or other important decisions.
• Availability and accuracy of AI features depend in part on third-party services and may be affected by rate limits, provider policies, or temporary unavailability.

8. Compliance and Responsible Use

• GrabDocs aims to use AI in a manner consistent with applicable privacy, security, and business requirements.
• GrabDocs considers operational, privacy, and security risks related to AI use and applies business rules, scoped retrieval, access controls, and related safeguards to help protect confidentiality and integrity.
• AI is used as an assistive capability intended to help users work more efficiently, not as a sole decision-maker for high-impact outcomes.
• GrabDocs reviews AI-enabled features and related controls over time to help support accuracy, security, and responsible use.

9. Logging and Audit

• Where applicable, GrabDocs logs AI-related activity (e.g., user, date, action, and feature usage) in line with its operational, security, support, and audit practices.
• Logs are designed to avoid storing full document content unless needed for debugging, security, or support.
• Logs are used for system integrity, troubleshooting, security, and support purposes and are not used to train AI models on customer data.

10. Reporting Issues

Users and institutions can report AI errors, misclassifications, bias concerns, or security/privacy issues by contacting:

• Support email: [email protected]

Reports are reviewed and addressed according to applicable internal support, security, and incident response processes.

11. Policy Updates

GrabDocs may update this AI Policy from time to time to reflect product changes, legal requirements, or evolving practices. Significant changes will be communicated as appropriate.

The latest version will be made available by GrabDocs through appropriate channels, such as customer documentation, direct notice, or a website posting if published there (including this page: GrabDocs AI Policy).

Continued use of GrabDocs after the effective date of an update may constitute acceptance of the revised policy, to the extent allowed under applicable terms.

This policy is part of GrabDocs' broader commitment to transparency, security, and responsible AI. For data protection and privacy details, see our Privacy Policy and your contract or data processing terms, if applicable.