Everything companies ask about the assessment.
Grouped by topic, answered plainly.

IThe assessment

What is the Humanistic AI Assessment?

An independent view of how your use of AI affects people. We look at the observable human consequences of how you adopt and deploy AI, score them across six domains, compare them with peers and show what to improve.

Who is a founding assessment for?

Companies where AI is central to the product or the way they work, or that are going through workforce change, have made public commitments on responsible AI, or invest in employee experience and employer brand.

How does it work?

In five steps. We survey your employees, interview management, collect evidence, cross-check the three sources against each other, then score and report. Where the sources disagree, we dig. See the full method.

What will you ask us for?

A 45-minute interview with your CEO or HR lead, a standard questionnaire, and 5 to 10 documents or data extracts, such as AI policies, training records, complaints and workforce metrics.

What do we get at the end?

A scorecard with overall and domain scores, the evidence gaps named plainly, and prioritised recommendations in plain language. Founding assessments also include a place in the founding benchmark and a reassessment a year later.

IIFramework and scoring

What are the six domains?

Dignity, Agency, Transparency, Responsibility, Fairness and Advancement, with one central question each. Privacy and security run through all of them. The framework is a working draft, refined with every founding assessment. See the framework.

How do you weigh the evidence?

What happens counts most. Observed outcomes are the strongest evidence, then operational records, then what employees and others tell us. Policies and published claims support a score but never earn one on their own.

Does having more policies raise our score?

No. Policies and commitments are supporting evidence only. More policies, committees or paperwork do not, by themselves, raise a score.

Who are you comparing us with?

Peers of similar industry, size and AI exposure. Benchmarks get sharper as the dataset grows with each assessment.

What if there is no evidence for something?

We separate missing evidence from evidence of poor performance, and say which is which. A gap is reported as a gap, not as a failure.

Can we see how a score was reached?

Yes. The framework, domains and method are public, and every company is scored on the same fixed rubric, so the same evidence should lead to the same score whoever does the assessment.

IIIPeople and privacy

Is the employee survey anonymous?

Yes. We run it, you never see individual answers, and results are only reported for groups large enough to protect identities.

Who else do you consider besides employees?

Five stakeholder groups: employees first, then customers, investors, partners and society. Each is affected by AI in different ways, so each is read through the six domains.

Where does AI come in, and where do people?

AI helps with research, evidence extraction, early analysis and drafting. People make material judgements, resolve disputes over evidence and review quality.

IVCertification and trust

Is this a certification?

Not on its own. We start with assessment and improvement. Certification is only for companies that score above the threshold, and it comes once the method is mature. It will never be presented as regulatory approval, and it is never a badge for sale.

What is the public trust page?

A public page for each certified company showing its scores, status, the evidence behind them and the date of its last assessment. You decide what is shown, and it updates when you are reassessed, so you can point people to one place instead of filling in the same questionnaires.

How long does certification last?

A year. Certification is renewed every year through reassessment, and the trust page shows the update history.

VIndependence and fit

How is this different from AI compliance or assurance?

Compliance, privacy, security and governance tell us a lot, and we use them as evidence. But we measure something else: the human consequences of how you use AI.

What don't you do?

We are not an AI implementation consultancy, and we don't do cybersecurity, technical AI testing or regulatory compliance consulting. We don't replace ISO, EU AI Act work or technical assurance.

How do you stay independent?

We don't sell implementation work, fees don't depend on results, and scores rest on evidence, not on how much a company pays.

Why the name Mirandola?

It comes from Giovanni Pico della Mirandola, who argued in 1486 that people are not fixed: we shape what we become. Good AI should give people more room to grow, not less. Read the founder's letter.

VIPrice and getting started

What does it cost?

Founding assessments are indicatively priced at £1,500 to £3,000, depending on size and scope. Founding pricing is still being tested and may change.

What is the founding benchmark?

We are assessing a small first group of companies. Their results, with names removed, become the first benchmark that later companies are compared against.

How do we start?

Leave your work email below. We'll send the draft framework, suggest a short call about where AI touches your people, and reply within a week. There is no obligation.

Still wondering?

Ask us directly.

No pitch. Your questions about the assessment, the evidence and whether it fits your company.

Book a founding assessment

Tell us a little about your organisation. We'll send the draft framework and suggest a short call about where AI touches your people.