AI Policy

What AI does in Metis Products

Metis products use AI in two distinct and separated layers:

• MKB Content Layer: AI is used as an assistive tool to help human researchers read, analyse and synthesise evidence from lawfully accessed sources into the Metis Knowledge Base (MKB). AI tools at this layer assist with source discovery, structural formatting, first-draft summarisation and thematic pattern identification. No personal user data is present or accessible at this layer.
• Application Layer: AI is used within Metis apps to match MKB content to your profile and workplace context, generating personalised outcomes. This is where your neurotype information, if provided, is used to tailor results to your cognitive profile. This processing occurs within the app, not within the MKB.

These two layers are kept technically and operationally separate. The MKB cannot identify you, and MKB content creation is not influenced by individual user data.

Neurotype-Aware Recommendation Logic

Metis app outputs are not generic. The MKB encodes neurotype-specific functional knowledge that enables Metis apps to filter and rank results based on the cognitive profile you provide. This operates at two levels.

At the inclusion level, Metis apps surface tools, strategies and approaches that align with your neurotype. At the exclusion level, which is equally important, Metis apps filter out results that are likely to be unsuitable for your profile regardless of how well-designed they are in general terms.

For example, for a user with Attention-Deficit/Hyperactivity Disorder (ADHD), Metis apps exclude tools and approaches that require heavy manual data capture, continuous logging, complex categorisation or regular review cycles to function effectively. If a tool depends on the user remembering to open it and update it consistently to deliver value, it is treated as a poor fit for that profile and will not be surfaced. This does not reflect a judgement about the overall quality of those tools; it is a recognition that the fit between a tool and a cognitive profile matters as much as the tool's general quality.

This filtering logic is grounded in the MKB's evidence layer, drawing on research into cognitive load, executive function, attention regulation and working memory as they manifest across specific neurotypes in workplace contexts. The result is that the outputs you receive have been screened for compatibility with your profile before you see them.

Sharing your neurotype is optional but may improve the quality of your results. Without it, Metis apps apply broader evidence-based guidance that is less tailored but still useful. The choice of what to share remains yours at all times, consistent with our core principle: 'Our story is our own'.

What AI does not do

It is equally important to be clear about what AI does not do within Metis products:

• AI does not provide professional advice: Metis AI outputs are general guidance only. They do not constitute medical, legal, occupational, psychological or clinical advice. You are responsible for exercising independent judgement before acting on any output.
• AI does not diagnose: Metis products do not diagnose health conditions, assess clinical presentations or prescribe treatment. Neurotype information is used to personalise outcomes only and is not a substitute for clinical assessment.
• AI does not train on your data: Metis does not use your personal information, neurotype data or usage patterns to train, build or fine-tune AI models. AI service providers engaged by Metis are contractually bound by the same restriction.
• AI does not train on third-party IP: Metis does not use third-party intellectual property, copyrighted source material or MKB content to train or build foundational AI models. This applies across internal workflows and contracted services.
• AI does not make governance decisions: AI tools do not make standalone compliance determinations, approve MKB content for publication or independently evaluate evidence tiers. These decisions remain with human reviewers.
• AI does not access personal data in MKB creation: Personal user data is not accessible to AI systems at the MKB content creation layer. The separation between the MKB layer and the application layer is a core architectural principle.

Medical and Clinical Limitations

Metis builds AI-powered assistive technology designed to reduce the invisible cognitive load of work for neurodivergent professionals. Built on the principle that 'Our story is our own', every Metis app is designed to respect your autonomy: you choose what to share, and you are not required to disclose a diagnosis to access support.

Sharing information about your neurotype is optional, but it can meaningfully improve the quality and relevance of the outcomes a Metis app provides. When neurotype information is entered, Metis apps use it to personalise results to your cognitive profile, including identifying approaches that are well-suited to how you work and filtering out those that are likely to create additional load rather than reduce it. The choice of what to share remains yours at all times.

Metis apps are not medical or therapeutic services. They do not diagnose health conditions, provide clinical advice or prescribe treatment. While Metis draws on peer-reviewed research, practitioner expertise and lived experience to support neurodivergent professionals in workplace contexts, our apps are not a substitute for qualified healthcare professionals. Decisions relating to diagnosis, treatment, medication or therapeutic support should be made in consultation with appropriately qualified practitioners.

Metis is designed to complement, not replace, professional medical, psychological or therapeutic care.

How AI is Governed

AI at Metis operates under a governance framework designed to keep human judgement at the centre of content decisions. The key controls are:

• Human-in-the-loop review: AI-assisted drafts for the MKB undergo human review and validation before inclusion. Reviewers check for residual source phrasing, structural replication, unlicensed content and hallucinated references. Drafts that do not meet transformation standards are edited or discarded.
• Prompt guardrails: Internal prompts are configured to prohibit reproduction of source expression and to direct AI tools toward extracting mechanisms, patterns and functional insights rather than original wording.
• Lawful source controls: AI tools at the MKB content layer process only material that has been confirmed as lawfully accessible. Sources obtained through unauthorised channels are not permitted.
• Audit trail: Metis aims to maintain records of AI-assisted drafting alongside human review logs to support ongoing compliance review.
• Accountability: The Content Lead is responsible for day-to-day AI compliance, including verifying that any new AI tool or service onboarded complies with the model training prohibition. The Founder / Director holds overall accountability.

Full operational detail on AI controls is set out in the MKB Governance Framework.

AI Outputs and your Responsibility

AI systems, including those used in Metis products, can produce inaccurate, incomplete or unexpected outputs. We do not guarantee that any output generated by a Metis product will be accurate, current, complete or suited to your specific circumstances. Outputs should be treated as a starting point for your own assessment, not as a definitive answer.

For any consequential decision, including those relating to workplace adjustments, professional development or health and wellbeing, we recommend seeking appropriate professional advice.

Your Rights and Controls

In relation to AI use within Metis products, you have the following rights and controls:

• Neurotype disclosure is optional: You are not required to provide neurotype information to access Metis products. Sharing it can improve the quality of outcomes you receive, but the choice is yours.
• Access to evidence sources: You may request general information about the evidence sources underpinning outputs you receive through a Metis app.
• Correction of personal data: You have the right to access and correct the personal information Metis holds about you, including neurotype data, in accordance with our Privacy Policy.
• Raising concerns: If you have concerns about how AI is used in a Metis product, contact us at info@metis.nz. If your concern relates to MKB content, contact mkb@metis.nz.

How this Policy evolves

AI technology and the regulatory environment around it are developing rapidly. In New Zealand, the Government has signalled a risk-based approach to AI governance, with existing consumer protection and privacy laws forming the primary compliance framework while sector-specific guidance continues to develop. Metis monitors these developments and updates its AI governance accordingly.

This Policy is reviewed at least annually and updated in response to material changes in AI technology, applicable law or our products and services. The most current version is always available at metis.nz.

Policy Administration and Review

This Policy is maintained by our Content Lead. For questions about AI use in Metis products, contact us at hello@metis.nz. For questions about MKB content and governance, contact mkb@metis.nz.