AI & Safety

How AXIS uses AI without pretending it is clinical judgement

What is automated, what is quality-checked, what cannot be inferred, and when AXIS is the wrong service.

Reviewed and updated July 21, 2026.

The live product boundary

Mirror uses no AI. Its four free self-reflection quizzes calculate results locally from fixed answer weights. The paid Diagnostic uses AI-assisted synthesis to turn your questionnaire answers into a structured Portrait. After that Portrait passes the required checks, AXIS renders the mobile 90-Day Companion deterministically from its validated content and generates the Complete Text Edition mechanically from the Portrait and Companion. Those two derivatives make no separate AI-model request and add no new AI interpretation, although they can carry through content from the AI-assisted Portrait. Coaching is a separate human-led service, available after a completed Diagnostic through a mutual-fit process.

There is no live AI coach or daily AI companion in the current service.

Controls on the Diagnostic

Evidence comes before interpretation
The system must ground observations in information you actually supplied, separate that evidence from interpretation, and leave gaps unfilled when the questionnaire does not support a conclusion.
Psychological language stays provisional
Developmental roots, attachment signals, protective functions, and core-wound language are working hypotheses for reflection. The report must state uncertainty, include plausible alternatives, and avoid claiming that a childhood event caused a present behaviour.
It does not diagnose or treat
AXIS does not diagnose depression, anxiety, PTSD, attachment disorders, personality disorders, or any other condition. It does not provide medical advice, treatment, or an assessment by a licensed clinician.
Disagreement is valid data
If an interpretation does not fit, you can reject it. The system may not treat disagreement as proof that you are defensive, resistant, unaware, or secretly confirming the claim.
Every Diagnostic must pass automated quality and safety checks
Before delivery, the AI-assisted Portrait is checked for unsupported certainty, clinical overreach, contradictions, duplicated passages, privacy issues, document completeness, and visual usability. The deterministic Companion is checked for source fidelity, completeness, and visual usability, and the Complete Text Edition is checked against both designed documents. Work that passes is released automatically; if any required check flags an issue, the complete delivery is held for review.
Sensitive data is used only to deliver the requested service
Your Diagnostic answers and documents are not used for marketing or AXIS model training. They are processed only by the providers needed to generate, store, review, and deliver your Diagnostic, under the terms described in the Privacy Policy.

How to read your report

Treat the portrait as a map to test, not a verdict about who you are. The useful question is not “is every sentence true?” but “which hypothesis is supported by my experience, what alternative also fits, and what small experiment would distinguish them?” Set aside anything that feels unsupported or harmful.

If you find a factual mistake, duplicated passage, privacy issue, unsafe claim, or interpretation presented with too much certainty, report it through your private Diagnostic page or email hello@axismethod.io. Every Diagnostic for which more than $0 was paid includes one correction pass requested within 30 days of first delivery for a material factual error or materially unsupported interpretation; discounted paid purchases are included and transactions totaling $0 are excluded. A qualifying correction regenerates the complete Diagnostic edition under the same quality and safety checks, preserves the original, and does not shorten the 30-day guarantee. It is not a style rewrite or a new questionnaire. Privacy and safety concerns can be reported regardless of correction-pass eligibility.

Feedback and testimonials

Seven days after delivery of your current Diagnostic, AXIS may send one non-promotional service-quality feedback request through Resend. If you request a correction before it is sent, the request follows the corrected edition instead. You can object or opt out through your private Diagnostic page. The request does not subscribe you to marketing, feedback is optional and private by default, and it asks you not to include new sensitive details. AXIS publishes only an exact quote for which you separately choose anonymous or named testimonial permission.

Crisis and clinical escalation

AXIS is not monitored crisis support and cannot respond to emergencies. If you are in immediate danger, thinking about suicide or self-harm, or experiencing a mental-health emergency, contact local emergency services or a crisis service in your country now. If you need diagnosis or treatment, contact a licensed mental-health professional.

Human coaching

After completing your Diagnostic, you may apply for coaching by email; no invitation is required. Coaching is human-led and remains educational self-development, not therapy. The Diagnostic establishes the working agenda, and you approve the target goals. Before any engagement, both sides confirm mutual fit and agree scope, privacy, call-summary choices, escalation boundaries, cancellation terms, and success measures. The Diagnostic remains complete on its own.

Questions
Something not covered
here?

If you have concerns about the Diagnostic, AI-assisted synthesis, privacy, or an unsafe interpretation, email us directly. We aim to respond within one business day.

hello@axismethod.io