AI Data Use Policy

Service Velocity · Kraken Services LLC, d/b/a Service Velocity Version: 1.0-draft Effective date: Upon publication (target: before V1 commercial launch, November 1, 2026) Applies to: All customers using AI-assisted features in the Service Velocity platform Incorporated into: Cloud Service Agreement

This AI Data Use Policy ("Policy") describes how Service Velocity ("Provider", "we", "us") collects, uses, retains, and protects data processed by artificial intelligence and machine learning features in the Service Velocity cloud platform ("Service").

This Policy supplements the Cloud Service Agreement, Data Processing Agreement (if applicable), and Acceptable Use Policy.

AI features include, without limitation:

• Drafting assistants — suggested email/SMS text, quote descriptions, job notes
• Workflow agents — automated actions within tenant-defined policies (dispatch suggestions, follow-ups, approvals)
• Search and summarization — natural-language queries over tenant data
• Document extraction — parsing uploaded files into structured records (where enabled)

Features that do not use third-party LLM inference (e.g., deterministic scheduling rules, geofencing) are outside this Policy unless they send content to an external model.

3.1 No training on your data

We do not use your Customer Content or Personal Data to train, fine-tune, or improve foundation models (ours or third parties'). Data is sent to LLM providers only to fulfill the specific request you or your Users initiate.

3.2 Tenant isolation

AI processing is scoped to your tenant (and workspace, where applicable). Models and prompts cannot access another customer's data. Outputs inherit the same access controls as the underlying data.

3.3 Human oversight

AI outputs may be inaccurate. You are responsible for reviewing customer-facing content before send, and for configuring approval thresholds for high-impact agent actions.

3.4 Opt-out and entitlements

AI capabilities are entitlement-gated by subscription plan. Tenant administrators may disable AI features or specific agents. Disabling AI does not delete historical usage ledger entries (see §5).

3.5 Policy-governed automation

Workflow agents operate only within tenant-configured policies: service areas, pricing guardrails, brand voice, approval thresholds, and dispatch rules. Agents cannot override explicit human denials where the product requires approval.

Depending on the feature, we may send:

We minimize context to what is reasonably necessary for the request. We do not send full database exports to LLMs.

5.1 Usage ledger

We record AI action events for billing, audit, and abuse prevention:

• Timestamp, tenant, user, feature/action type
• Token counts (where available from provider)
• Outcome status (success, failure, blocked by policy)

5.2 Prompt and completion content

Do not publish a hard 90-day deletion claim until the purge job exists. Until then, customer-facing language should describe DB-persisted conversation history plus operational log retention.

We do not retain prompt/completion content for model training.

5.3 Aggregated analytics

We may use aggregated, de-identified statistics (e.g., feature adoption rates, average tokens per action) to improve product reliability and pricing. Aggregates cannot reasonably identify your business or end customers.

AI inference is performed by subprocessors listed in the Subprocessor Exhibit, primarily Microsoft Azure OpenAI Service. Provider contractually restricts subprocessors from using Customer data for training except as required to provide the inference API and as permitted by this Policy.

If you process EU/UK personal data, our DPA and Standard Contractual Clauses govern transfers to U.S.-based AI subprocessors.

AI API calls use encrypted transport (TLS). Credentials for LLM providers are stored in Azure Key Vault. Tenant-scoped authorization is enforced before any AI request is executed.

You agree to:

• Ensure you have rights to submit Customer Content for AI processing
• Not submit prohibited data (payment card numbers, government IDs, health records, etc.) into AI prompts unless the Service explicitly supports that use case
• Configure approval workflows appropriate to your risk tolerance
• Inform your Users that AI features are assistive, not professional advice (legal, tax, safety, licensing)

We may update this Policy with 30 days' notice for material changes (email to account administrators and updated posting at /legal/ai-data-use). Continued use of AI features after the effective date constitutes acceptance.

AI & privacy questions: privacy@servicevelocity.ai Security incidents: security@servicevelocity.ai

*Draft for counsel review — not yet published for customer acceptance.*