Legal

How we use AI and human review

Effective date: 10 October 2026. Version 0.8.

We use artificial intelligence to help build your artefact, and we think you should know exactly how, before you buy rather than after.

This page is the full account. There is a one-line version on the sales page and in the intake form, and it links here.

1. The short version

Your artefact is drafted with the assistance of an artificial intelligence model, working from a set of instructions we wrote, using only the information you gave us. Robert Russell reads it before it is sent to you. No artefact is delivered without that review.

2. What the AI actually does

The AI model receives a structured instruction we have written and refined, plus your information. It produces the draft text of your artefact: the analysis, the framing, the observations and the guidance.

It does not choose your price, it does not decide whether you are accepted as a customer, it does not score or rank you against other people, and it does not decide anything about you that has a legal or similarly significant effect. It writes a draft of a document.

3. What information the AI sees

It receives exactly this and nothing else:

Your first name
Your surname, for a comprehensive build only. Eigenprint Name traces your full name to its roots, so it needs your surname as well as your first name
Which of the three starting points on the form describes you
Your date of birth
Your sex, gender, religion and ethnicity
Your Eigenprint Type result
Your four written career answers
Your seven answers about your working setup, how steady things feel for you and what your work means to you
Your answers to the written questions we send you after you pay for your artefact

It never receives:

Your individual answers to the Eigenprint Type assessment. The model receives the result those answers produce, never the answers themselves
Your email address
Your telephone number
Your company
Your location
Your payment details
Your account details

This is deliberate. The model needs your first name to address you, your surname for the comprehensive build so that Eigenprint Name can trace your full name, your birth date and your type result to work out your results, your starting point to write for where you actually are, your sex, gender, religion and ethnicity because they shape the advice and the roadmap, and your answers to say anything useful. Religion and ethnicity are special personal information, so you agree to them being sent to the model in the consent you tick before you pay. Everything else would add nothing to your artefact and would only spread your information further, so we leave it out.

Your individual assessment answers stay out for the same reason. We keep them, because a result you were given should always be checkable against what you actually said, but the model only ever needs the result.

4. Who provides the model

Anthropic, through its application programming interface. Anthropic processes the information described in clause 3 on our instruction, under a written data processing agreement, in the United States.

Your information is not used to train any AI model. Training is how an AI model learns from the material it is given. Our contract with Anthropic does not allow it to train a model on your intake answers, your answers to the written questions or your artefact.

Anthropic deletes the information we send within 30 days of receiving it. Anthropic also deletes the draft its model writes within 30 days. There are only two cases where Anthropic keeps them longer:

If the law requires Anthropic to keep them, Anthropic keeps them for as long as the law says.
If Anthropic’s safety checks pick up possible misuse of its service, Anthropic can keep them for up to two years. Misuse includes an attempt to use the model to cause harm.

We may change the model or the provider as the technology moves. If we do, we update this page and the sub-processor list in our Privacy notice.

5. What the human does

Robert Russell reads every artefact before it is sent. This is not a formality and it is not a spell check. The reviewer is checking for:

Data accuracy. That your name is spelled correctly, that your birth date was read correctly, that the Eigenprint Path calculation is right and shown in its correct form, and that the Eigenprint Sky follows from the date.
Framework accuracy. That the frameworks are described as they actually are, and that nothing has been invented or blended.
Attribution. That nothing is presented as your own words unless you actually wrote it.
Tone. That the document is honest rather than flattering. Generic, over-spiritualised, affirmation-heavy writing is the specific failure we watch for, and it is the reason the review exists.
Harm. That nothing in it is inappropriate, that it stays inside what we do, and that it does not stray into medical, psychological, financial or legal territory.

The reviewer can pass the document, amend it, or hold it and have it rebuilt. Every review is logged with the outcome, the reviewer and the time.

This is why a build takes working days rather than minutes. An instant artefact is an unreviewed artefact. The time it takes is the quality control.

6. The statement that appears in your artefact

Every artefact carries this, on the inside front page:

This document was drafted with the assistance of an artificial intelligence model, working from information you provided, and was reviewed by Robert Russell of QuantumThrive Hub before it was sent to you. It is guidance, not advice, and not a psychological or psychometric assessment. If anything in it is factually wrong, write to support@quantumthrivehub.com and we will correct it and rebuild the affected part at no charge.

7. What this means for the quality of your artefact

Being straight about the limits is part of the product.

AI writing can be confidently wrong. That is why a person checks it, and why we ask you to check the facts about yourself and tell us if anything is off.
AI writing tends towards the generic and the flattering. We have written the instructions specifically to resist that, and the reviewer’s main job is to catch it when it happens anyway.
The frameworks are interpretive. Eigenprint Type, Eigenprint Path and Eigenprint Sky, and in the comprehensive build Eigenprint Name and the timing work, are lenses for thinking about yourself. They are not measurements, and they are not predictions. Our Disclaimer and scope of service goes into this properly.
The quality of your artefact depends heavily on the quality of your answers. Four thin answers produce a thin document. This is the one part of the process we cannot do for you.

8. Your rights in relation to automated processing

Because a person reviews every artefact, we do not consider that we make decisions about you by automated means alone. Even so, you are entitled to:

Ask us how your artefact was produced and what the AI contributed.
Tell us you disagree with something in it.
Ask a person to look at it again.
Ask us to correct it.

Email support@quantumthrivehub.com. We respond within two business days.

9. Where else we use AI

Where	What it does	Human involvement
Eigenprint Type taster	Not used. Your taster is fixed, pre-written interpretation text, selected by the scoring service from your assessment result. No AI model drafts it	Not applicable. There is nothing an AI model wrote for a person to review
Artefact generation	Drafts your artefact	Full review of every artefact before delivery
Internal drafting	Helps us write marketing copy and internal documents	Every published word is written or approved by a person
Support correspondence	Not used. Replies from support@ are written by a person	Entirely human

We do not use an AI chatbot to talk to you. If you get a reply from us, a person wrote it.

10. Regulatory note

Article 50 of the European Union Artificial Intelligence Act introduces transparency obligations for certain AI systems and AI-generated content, applying from 2 August 2026. Our funnel does not deploy a conversational AI system, and your artefact is a personal document rather than content published to inform the public. We nonetheless disclose our use of AI clearly, before purchase, in the intake, in the delivered document and here, because that is the standard we want to be held to regardless of what is strictly required.

11. Revision history

Version	Date	Change	Approved by
0.1	31 July 2026	First draft for review. Not published	Pending
0.2	25 September 2026	Supplier details filled for a sole proprietor, retention and refund rulings applied, Path and Sky described by how they work	Robert Russell
0.3	2 October 2026	The free scorecard result row comes out of section 9. There is no free scorecard, and no AI assembles the Eigenprint Type taster	Robert Russell
0.4	2 October 2026	Section 9 gains an explicit Not used row for the Eigenprint Type taster, rather than leaving the free thing undisclosed by omission, matching the existing row for support correspondence	Robert Russell
0.5	5 October 2026	Section 3 now lists the answers to the written questions we send after you pay for your artefact. Section 4 now explains what training an AI model means. That section also says how long Anthropic keeps the information we send and the draft its model writes	Robert Russell
0.6	8 October 2026	Section 3 now lists your sex, gender, religion and ethnicity among the answers the model receives, in line with the checkout consent	Robert Russell
0.7	11 October 2026	Sections 1, 5 and 6 say Robert Russell reads every artefact, as the terms of sale do. Section 7 names Eigenprint Name and the timing work in the comprehensive build. Other documents are called by their page headings	Robert Russell, 15 October 2026
0.8	16 October 2026	Section 3 moves your surname from the list of what the model never receives to the list of what it receives, for a comprehensive build only	Robert Russell, 16 October 2026