AI Trust & Controls
Inspire uses AI to reduce admin work and surface insights — not to make decisions about your employees. This page explains exactly what data AI can access, what it cannot touch, how it's controlled, and how humans stay in the loop at every step.
Last updated: February 2026Reviewed by: Inspire Product & Security Team
What data can Inspire's AI access?
AI CAN access
- OKR titles, descriptions, progress percentages, and confidence scores — within the acting user's permissions
- Check-in notes and responses that the acting user submitted or can view
- Goal status and milestone history for the user's direct reports
- Recognition messages visible to the acting user
- Review status (not content) for administrative scheduling purposes
- Aggregated team metrics (completion rates, engagement rates) for manager dashboards
AI CANNOT access
- Data outside the acting user's role-based permissions (AI never bypasses access controls)
- Social Security numbers, financial data, health records, or sensitive PII
- Performance data of employees the acting user doesn't manage
- Raw review content from other managers' direct reports
- Calendar data, emails, or communications outside of Inspire
- Historical AI-generated outputs (each AI call is stateless)
How is AI-related data retained?
1AI suggestion accepted and saved
Becomes part of the standard platform record. Subject to your normal data retention settings (typically until contract end + 30 days grace period).
2AI suggestion rejected or ignored
Not stored. The suggestion is discarded immediately when the user exits without saving.
3Raw AI API call data
Not logged in Inspire's systems beyond the immediate processing window. AI provider API logs follow the provider's standard data handling terms (details available in our DPA).
4AI model training
Never. Your data is not used to train Inspire's AI models or any third-party models used by Inspire. This is a hard contractual and technical boundary.
How do admins control AI features?
Organization-level toggle
Admins can enable or disable all AI features for the entire organization from the admin panel.
Feature-level controls
Individual AI features (agenda generation, review drafts, trend insights) can be enabled or disabled independently.
Manager opt-out
Individual managers can choose not to use AI suggestions without affecting other users — no admin action needed.
Always draft, never auto-submit
AI-generated content requires explicit user action to save. Nothing is submitted automatically. Every review, check-in note, and recognition post is human-initiated.
What is Inspire's PII stance for AI features?
Inspire’s AI features are designed to work with work-related performance context, not personal identity information. Specifically:
- •AI processes employee names and roles only as context labels — not as identity verification data
- •No financial, health, demographic, or biometric data is processed by AI features
- •AI prompts are constructed to use pseudonymized data wherever functionally possible
- •Customer data sent to AI APIs is subject to zero-data-retention API agreements with our providers
What are the AI model boundaries and training policies?
Does Inspire fine-tune models on customer data?
No. We do not fine-tune or train any models using customer data. AI models are used via API — your data is never used for model improvement.
Can AI hallucinate incorrect employee information?
AI features in Inspire operate on structured data from the platform — they don't retrieve information from the web or external sources. This significantly reduces hallucination risk for the use cases we support. That said, all AI output should be reviewed before use.
What models are used?
Inspire uses leading AI APIs from enterprise-grade providers. Specific model details are disclosed in our enterprise DPA. We evaluate and update models as better options become available — security terms are maintained throughout.
AI SAFETY
What is Inspire's AI safety posture?
No autonomous scoring
AI does not generate employee performance scores, risk ratings, or pay recommendations. Advisory only.
Explainable suggestions
AI suggestions include context about the data they're based on. No black-box outputs without reasoning.
Human in the loop - always
Every AI output requires a human action to take effect. There is no autonomous AI decision-making in Inspire.
Auditability
All saved content (including AI-assisted content) is logged with user attribution and timestamps. Admins can audit who saved what, when.
AI Trust FAQs
Does Inspire train AI models on my company's data?
No. Your company data is never used to train Inspire’s underlying AI models or any third-party AI models. AI features process your data in-context to generate outputs (agendas, summaries, drafts) but do not persist that data for model training purposes.
Can employees or managers opt out of AI-generated content?
Yes. Administrators can control which AI features are enabled at the organization level. Individual managers can choose to disregard or rewrite any AI-generated content before submission. AI outputs are always drafts — never automatic submissions.
Can AI see employee PII?
AI features have access to work-related performance data (OKR progress, check-in notes, goal status) within the permissions of the acting user. AI does not access financial data, Social Security numbers, health records, or other sensitive PII beyond what’s in employee work records.
What AI provider does Inspire use?
Inspire uses leading AI APIs to power AI-assisted features. Specific provider details are available in our security documentation and DPA for enterprise customers. All AI provider agreements include data processing terms consistent with our customer commitments.
How long is AI-generated content retained?
AI-generated content (e.g., agenda suggestions) that is accepted and saved becomes part of the regular platform record with standard data retention rules. Content that is rejected or ignored is not stored. Raw AI API calls are not logged beyond immediate processing.
Can AI make decisions about employee performance or compensation?
No. AI in Inspire is advisory only. It drafts, summarizes, and suggests — it does not rate, rank, flag, or make decisions about employees. All review submissions, ratings, and recommendations are made by humans and explicitly require human action.
Is AI in Inspire explainable?
Yes, within scope. AI-generated suggestions include context about what they’re based on (e.g., ‘Based on your last 3 check-ins and current OKR confidence scores…’). Trend indicators include the data signals that drove them. We don’t surface AI outputs without context.
How are role-based permissions applied to AI features?
AI features respect the same role-based access controls as the rest of the platform. A manager’s AI-assisted review draft only includes data the manager has access to. Administrators cannot use AI to access data outside their permissions. AI does not bypass any existing access rules.
