AI in assessment
requires a standard.

Assessment affects
people's careers.

Competency-based assessment reports inform promotion decisions, development investments, and leadership pipeline planning. The people being assessed are not data points — they are professionals whose working lives are shaped, in part, by the quality and integrity of the assessment process.

When AI enters that process, the stakes of getting it wrong rise. An AI that invents observations, softens unfavourable findings, or introduces bias it was never instructed to introduce can cause real harm to real people — and to the practitioners and firms whose reputations depend on the integrity of their work.

The Guild Standard exists because we believe AI in assessment must be held to explicit, published commitments — not left to assumption.

What The Guild AI does.

Every AI feature in The Guild operates within these boundaries:

1. 1.Enhances written language — grammar, clarity, professional tone, and sentence structure — in assessor-written narratives.
2. 2.Transcribes spoken assessor observations accurately and organises them by competency (Scribe).
3. 3.Validates report structure against the practitioner's uploaded competency framework and template.
4. 4.Flags misalignments between scores and narratives, missing sections, and format inconsistencies.
5. 5.Presents all suggested changes transparently, with every suggestion visible and reversible.

What it does not do.

1. 1.Invent, infer, or fabricate observations that were not recorded by the assessor. If the assessor did not observe it, The Guild will not write it.
2. 2.Change scores, ratings, or competency assessments. Scores are set by the assessor. They are not touched by any AI feature.
3. 3.Generate development recommendations without an assessor observation to ground them.
4. 4.Remove unfavourable findings or soften critical feedback without explicit assessor instruction.
5. 5.Access, process, or store candidate data beyond what is required to generate the report in session.
6. 6.Make final decisions. The Guild presents. The practitioner decides. Always.

The assessor's judgment
is always final.

Every AI suggestion in The Guild is a suggestion. The assessor sees what changed, why it changed, and can accept, modify, or reject every edit — individually, not in bulk. No AI output leaves The Guild without a human practitioner having reviewed and approved it. This is not a feature. It is a principle we will not compromise.

Our commitment
to the profession.

We commit to updating The Guild Standard whenever we introduce a new AI capability — before that capability is released, not after.

We commit to being transparent about the limitations of our AI — including when it gets things wrong and how practitioners should check its output.

We commit to never using assessment data processed through The Guild to train AI models without explicit, informed consent from the practitioner and their client.