lumi — AI Insights Terms

Version 1.1 · Effective 3 September 2026

What "AI Insights" are

AI Insights are short, written interpretations of your own benchmark figures — for example a plain-English read of how a reward domain sits against your peer group, or commentary on a single metric. They are generated by a large language model and are descriptions of your data, not advice. They never introduce figures that aren't already on your screen, and every output passes an automated grounding check before you see it; if it fails, a rule-based summary built only from your numbers is shown instead. Each insight is shown with a notice that it is AI-generated.

What data is used

To generate an insight, lumi sends the model only already-computed, aggregated and derived figures for the relevant view — domain-level position counts, named benchmark metrics with their percentile and sample-size figures, prevalence figures, and the size of your peer comparison group. Individual salaries and base-pay levels are not used. We never send raw submissions, another organisation's data, or personal data about individuals beyond the organisation-level figures you have entered.

The AI provider (sub-processor)

Insights are generated by Anthropic PBC, our AI sub-processor, under a data-processing agreement. Anthropic processes the figures on lumi's instructions to return the generated text and, under its commercial API terms, does not retain the data or use it to train its models. Anthropic is named in the Sub-processor List with the processing purpose and safeguards.

Lawful basis and your control

We process these figures to provide AI Insights on the basis of our legitimate interest in offering analytical features within the service (UK GDPR Art. 6(1)(f)); a Legitimate Interests Assessment is on file. Because the figures are organisational and aggregated, and the output is descriptive only, this processing is proportionate and low-risk.

• AI Insights are on by default as part of your analytical experience.

• You can turn AI Insights off at any time in Settings. Doing so stops any further AI processing of your data for insights immediately.

• This setting is per person — each member decides for themselves. An Admin may additionally switch AI Insights off for the whole organisation, which overrides individual settings.

• AI Insights produce descriptive text only and make no automated decision with a legal or similarly significant effect (UK GDPR Art. 22 does not apply).

Accuracy

AI-generated summaries are produced automatically and may contain errors. They are a description of your data, not advice, and must not be relied on as a substitute for professional reward, legal, financial or HR advice. You remain responsible for all decisions about your reward arrangements. Where the AI service is unavailable or its output cannot be verified, lumi shows a plain rule-based summary instead.

Changes

If we make a material change to how AI Insights work, we may ask you to review the updated terms. The version in force when your setting was recorded is kept with the date.

---

See the Privacy Notice and Sub-processor List for related detail.