Security, privacy and responsible AI by design
GrundMind protects participant work and client information through restricted assessment access, evidence-aware analysis, immutable result records, human-controlled publication and controlled report delivery.
- Secure assessment access
- Protected, immutable evidence
- Human review and publication
- Controlled report delivery
Access narrows at every stage. No participant sees another participant, and client findings are delivered as a controlled report rather than platform access.
What we commit to
Restricted participant access
Participants access only the survey assigned to them through an individual secure link. They cannot access other participants, administrative functions, results or reports.
Controlled client reporting
Client findings are currently delivered through a controlled reporting process rather than direct platform access. Reports are prepared from protected result records and delivered only to authorised client representatives.
Evidence-aware assessment
GrundMind applies defined evidence requirements before displaying results. Missing information is never treated as a negative score, and results are suppressed when the available evidence is insufficient.
Human-controlled publication
AI may support structured interpretation, but it cannot automatically produce a client-facing result. Narratives must be reviewed and explicitly published by an authorised administrator before they appear in a report.
What is public, what is controlled, and what remains deployment-specific
DPIA support
Prepared · controlledGrundMind has prepared product-level data-flow, risk and safeguard documentation. Where a customer is the controller, that material can support the customer's own DPIA assessment where one is required.
Controller / processor roles
Documented · engagement-specificGrundMind has documented its default controller/processor position, while preserving the requirement to determine roles from the purposes and means of the actual deployment.
GDPR / Swiss FADP
Controls documentedPrivacy, access, minimisation, lifecycle and human-oversight safeguards are built into the service and documented in the Trust Pack. Legal compliance still depends on the deployment, lawful basis and responsibilities of each party.
EU AI Act position
External review plannedGrundMind's intended-use boundary excludes employment decision-making and automated work allocation. This is a product design position, not a formal external legal opinion or compliance certification.
References to GDPR, the Swiss FADP, the EU AI Act or ISO standards describe intended safeguards, design references and current readiness work. They are not claims of formal legal or regulatory compliance, certification, or external legal classification unless explicitly stated.
What procurement can request
- Trust & Compliance OverviewPublic
Executive overview of GrundMind security, privacy, responsible-AI and current assurance status.
View public PDF - Responsible AI & Acceptable Use StandardPublic
Published intended-use, prohibited-use, human-oversight and AI-assistance boundaries.
View public PDF - Privacy PolicyPublic
Public privacy information for website and service use.
- Technical & Organisational Measures (TOMs)Controlled
Prepared security-control schedule; shared selectively during qualified due diligence.
- Data Processing Agreement (DPA)Controlled
Working contract template prepared; engagement schedules and targeted legal review remain required before signature.
- DPIA Support PackControlled
Prepared product-level risk, data-flow and safeguard material to support a customer/controller DPIA where applicable.
- Controller / Processor Position PaperControlled
Prepared role analysis; the final controller/processor position remains engagement-specific.
- Subprocessor & Hosting RegisterControlled
Prepared partially verified supplier/hosting register; several production account and location facts remain explicitly open.
- Retention & Deletion StandardApproval draft
Prepared lifecycle and proposed retention schedule; owner approval is required before proposed periods become adopted policy.
- Data Flow & Architecture OverviewControlled
Prepared logical assessment-to-report architecture, access boundaries and reporting-layer separation.
- Security & Incident Response OverviewControlled
Procedure prepared; named ownership and the first tabletop exercise remain pending.
- Security questionnaire supportControlled
Procurement responses are maintained in an internal answer bank and supplied selectively; the internal master is not distributed wholesale.
- EU AI Act intended-use / classification positionExternal review planned
Current position reflects product design, restricted use and human oversight; external legal confirmation is planned before any formal classification claim.
Detailed security, risk, architecture and legal-review material is not published indiscriminately. Controlled information may be provided to qualified customers during due diligence and under confidentiality arrangements where appropriate.
Prefer email? Contact security@grundmind.com.
Standards references and assurance status
ISO/IEC 27001:2022
Control mapping in progressGrundMind maintains a documented security-control set and is mapping it against ISO/IEC 27001:2022. This remains an internal alignment activity and is not an ISO certification claim.
Swiss FADP and GDPR
Controls documented · deployment-specific reviewPrivacy, access, controller/processor, DPA and DPIA-support documentation is now available. Applicable lawful basis, deployment obligations and customer responsibilities remain engagement-specific; this is not a blanket compliance claim.
EU AI Act intended-use position
Internal position · external review plannedGrundMind is designed for AI adoption support, workflow improvement, enablement, governance and development — not employment decision-making. The current regulatory position is an internal product interpretation, not an external legal classification opinion.
OWASP application security guidance
Application controls reviewedApplication security controls are reviewed with reference to recognised OWASP guidance as the platform and deployment model evolve.
Responsible AI governance
Public standard publishedGrundMind's Responsible AI & Acceptable Use Standard now documents intended and prohibited uses, human oversight, deterministic scoring boundaries, evidence traceability and controlled publication. This is not an ISO/IEC 42001 certification claim.
SOC 2
Not claimedGrundMind does not currently claim a SOC 2 Type I or Type II assurance report. Assurance reports held by infrastructure providers do not transfer SOC 2 assurance to GrundMind itself.
References to standards describe GrundMind's internal control alignment and assessment activities. They do not represent independent certification unless explicitly stated.
From invitation to report
- 1
Secure invitation
Each participant receives an individual link to the survey assigned to them.
- 2
Restricted assessment access
The participant can access only their assigned assessment. They cannot access reports, administrative functionality or another participant's information.
- 3
Durable response saving
Responses are saved progressively throughout the assessment. Participants can resume their work after closing the browser or changing sessions.
- 4
Protected submission
Save failures are shown visibly rather than being falsely presented as successful. Final submission is protected against duplicate completion and linked to the exact survey version completed by the participant.
- 5
Evidence-aware analysis
Results are generated only when defined evidence requirements are met. Participant, team, manager and organisational evidence remain separated throughout the analysis.
- 6
Human-reviewed reporting
Analysis is reviewed, explicitly published and rendered into a controlled client-facing PDF for authorised recipients.
How we protect assessment data
Security controls are reviewed and improved as the GrundMind platform, methodology and client requirements evolve.
Safeguards built into the assessment and reporting workflow
Durable participant submissions
Responses are saved progressively and can be resumed securely after closing the browser or changing sessions. Save failures remain visible, recovery is handled safely after reconnection and final submission is protected against duplicate completion.
Immutable result records
Individual, Team, Manager, Manager-Team Alignment and Corporate Results are generated as immutable snapshots. Their source campaign, team, evidence, sample size, generation time and report version remain retained.
Evidence thresholds
Scores are shown only when defined numeric-response and subdimension-coverage requirements are satisfied. Dimensions with insufficient evidence are suppressed, and missing evidence is never interpreted as zero.
Separated reporting layers
Participant evidence, Individual Results, employee Team Results, manager perspectives, Manager-Team Alignment and Corporate Results remain distinct. Manager and employee evidence are not blended into a misleading common average.
Human-controlled publication
AI may support structured interpretation, but client-facing narratives require human review and explicit publication. Editable working drafts remain separate from the published narrative included in reports.
Lifecycle preservation
Archived organizations are removed from normal operational views while their submissions, results, reports and provenance remain protected for authorised historical access. New result generation is blocked for archived organizations.
Evidence integrity by design
GrundMind does not calculate a result from whatever information happens to be available. Defined evidence rules determine whether a result can be shown and how confidently it can be interpreted.
Minimum numeric evidence
A defined minimum number of scored responses must be available before a dimension can be reported.
Subdimension coverage
Evidence must cover enough of the underlying subdimensions. A score based on a narrow portion of the assessment is not presented as complete.
Minimum reporting groups
Team-level results are produced only when the applicable minimum reporting-group requirement is satisfied.
Transparent evidence status
Results distinguish between full evidence, partial evidence and insufficient evidence.
Suppression rather than invention
A result is withheld when the available evidence is insufficient. The platform does not manufacture certainty from incomplete responses.
Missing does not mean zero
Questions that are unanswered or not applicable are not treated as negative performance or low capability.
GrundMind reports the limits of the available evidence instead of converting missing information into misleading precision.
Distinct reporting layers
Different sources of evidence answer different questions. GrundMind preserves those distinctions throughout the assessment and reporting process.
- 1Participant source evidence
- 2Individual Results
- 3Employee Team Results
- 4Manager perspective
- 5Manager's personal Cognitive Fit
- 6Manager-Team Alignment
- 7Corporate Results
- 8Reviewed interpretation and recommendations
Employee and manager evidence are not blended into one common score. Manager-Team differences are retained as diagnostic signals and do not establish that either perspective is automatically correct.
Participant privacy
GrundMind assessments may contain information about employees, teams, workflows and organisational practices. Privacy safeguards are applied throughout invitation, participation, analysis, reporting, archival and deletion. The customer will often determine the purpose of an employee assessment, but the exact controller and processor roles must be defined from the facts of each engagement rather than assumed.
Read the GrundMind Privacy Policy- Participants receive privacy information before submitting an assessment.
- GrundMind collects only information required for the agreed purpose.
- Access to identifiable responses is limited to authorised personnel.
- Source-answer inspection is restricted to administrators.
- Clients are informed about how individual and aggregated results will be presented.
- Retention and deletion are managed under defined lifecycle principles; specific retention periods remain subject to the approved policy and engagement requirements.
- Participants can request information about how their personal data is processed.
- Personal data is not sold.
- Client or participant data is not used to train public AI models.
- Subprocessor use is documented and subject to contractual and data-protection review appropriate to the processing.
Responsible AI use
The methodology is human-designed; AI assists only with draft interpretation.
The GrundMind methodology itself is not an AI tool. It is a human-designed diagnostic method based on structured surveys, predefined scoring rules, evidence thresholds and reporting layers.
However, the GrundMind platform contains an AI-assisted component because AI helps draft interpretations or narratives from the calculated results. That being said, AI is used only to support draft interpretation, which must be reviewed and explicitly published by an authorised human before client delivery.
Deterministic result generation
Core assessment results are generated through defined scoring and evidence rules rather than unrestricted AI interpretation.
Human-controlled publication
AI-assisted content requires review, editing where necessary and explicit publication before client delivery.
Evidence transparency
Reports distinguish source evidence, calculated results, interpretation and recommendations.
Separated perspectives
Participant, employee, manager and organisational perspectives remain distinct rather than being combined into misleading averages.
No automated employment decisions
GrundMind is not designed to independently determine recruitment, dismissal, promotion, compensation, disciplinary action or other employment outcomes.
From evidence to a published report
- Immutable evidence
- Deterministic result
- Analysis brief
- Editable reviewed narrative
- Saved draft
- Explicit publication
- Controlled PDF
AI-assisted interpretation does not automatically become part of a client report. An authorised administrator reviews the analysis, edits the narrative where required and explicitly publishes the approved version.
Published narratives are retained separately from editable drafts so the wording delivered to the client remains traceable.
GrundMind is intended for organisational AI adoption, workflow improvement, enablement, governance, capability development and developmental support. It is not designed or permitted for individual employment decision-making. GrundMind outputs must not be used to make, recommend, rank, score or materially influence decisions concerning an individual’s recruitment or hiring, promotion, compensation, performance evaluation, disciplinary action, termination, selection for redundancy or headcount reduction, or allocation of work, tasks, shifts or responsibilities.
GrundMind may identify organisational, workflow, capability or structural changes for consideration, but its outputs must not be used to determine which individuals should be selected for employment-related actions. All findings require human interpretation and may be used only within GrundMind’s stated organisational and developmental purposes.
Traceable reporting
Each generated result retains the context needed to understand what it represents and where it came from.
A delivered report is based on a defined result snapshot. It is not a live calculation that silently changes when source data or interpretation is later updated.
- Result type
- Source organization
- Source campaign
- Source team
- Included evidence
- Sample size
- Evidence status
- Survey version
- Generation timestamp
- Result version
- Published narrative version
- Report version
Controlled client-ready reports
GrundMind generates reports through a dedicated report-only rendering process designed to separate client-facing content from internal administrative and analytical material.
What a client receives is a controlled report, not a screenshot or unrestricted view of the underlying platform.
- Administrative navigation is excluded
- Internal controls are excluded
- Internal analyst material is excluded
- Immutable result evidence is used
- Only the published narrative version is included
- Team, Manager and Corporate reports use dedicated client-ready formats
- A4 page-break handling supports professional PDF output
- Reports are delivered only to authorised recipients
Controlled information lifecycle
GrundMind supports controlled archival without silently destroying the evidence behind completed assessments.
Active organizations
- Included in normal operational dashboards
- Included in current warnings and readiness indicators
- Eligible for new result generation
- Available for active campaign and reporting workflows
Archived organizations
- Removed from normal operational warnings and counts
- Available through dedicated archived views
- Historical submissions and results retained
- Existing reports and provenance preserved
- New result generation prevented
- Restorable through a controlled administrative action
Retention, archival and deletion obligations may differ by engagement. GrundMind applies the agreed contractual and legal requirements for each client.
Operational oversight
Authorised administrators can identify incomplete or blocked assessment and reporting workflows without accessing the platform through participant-facing links.
Operational visibility helps identify missing evidence, incomplete result generation and delivery issues before client reporting is finalised.
Intended use of GrundMind
GrundMind is an organisational AI adoption and work-fit diagnostic and advisory platform. It is intended for AI adoption support, workflow improvement, enablement, governance and development. Results require human interpretation and should not be used as the sole basis for decisions about an individual.
GrundMind is not intended to:
- Make hiring or recruitment decisions
- Determine promotion, compensation, disciplinary action or termination
- Rank employees or managers for employment purposes
- Serve as a performance-evaluation system
- Automate the allocation of work based on an individual's profile or score
- Replace professional HR, legal, regulatory or management judgement
- Serve as a GxP system of record
- Maintain FDA-regulated electronic records or execute regulated electronic signatures
- Make clinical, manufacturing, product-release or patient-safety decisions
Trust is built into the assessment workflow
GrundMind is designed to help organisations adopt AI responsibly. That responsibility extends to how participant work is saved, how evidence is evaluated, how perspectives remain separated and how findings are reviewed, published and delivered.