Governance framework & guidance
What is an AI Use assessment?
Assessment asks what a particular organisational use of AI means for the organisation and what governance attention it requires. It considers facts such as purpose, people affected, information involved, reliance on AI output, significant decisions, potential consequences, human oversight, product suitability and unresolved uncertainty.
The purpose is not to prove that an AI system is completely safe or reduce the Use to one score. It is to understand the particular use well enough to make proportionate, defensible governance decisions.
Why AI should be assessed in context
The same AI Product might brainstorm social-event ideas, summarise sensitive employee information, or help rank employment applicants. The product can be identical while the consequences, data, people affected and human oversight are very different.
Swell therefore assesses an organisational AI Use, with product and supplier information informing the assessment rather than determining it on its own.
Ask for facts, not specialist jargon
Swell is designed around a simple principle: ask people for facts they can reasonably know and do not ask them to make specialist judgements they are not qualified to make.
A contributor can say “we use this tool to summarise customer complaints before a staff member reviews them”. They should not need to decide that this represents a particular technical risk category. Swell’s rules and specialist reviewers handle the governance translation.
Unknown is not the same as No
If a contributor cannot establish an answer, they should not choose the answer that looks safest merely to complete the assessment. Swell preserves “I don’t know” and “Needs confirmation” so uncertainty remains visible until it is resolved.
Treating unknown information as a favourable “No” would create false assurance.
Proportionate assessment
Not every AI Use needs the same depth of assessment. A straightforward internal drafting use should not automatically receive the same scrutiny as an AI system influencing employment or customer decisions.
Swell uses progressive disclosure so simple uses remain relatively quick while greater impact or uncertainty triggers additional questions and specialist review.
Using Swell AI
How Swell determines findings
Swell’s core governance assessment is deterministic. Defined facts and conditions trigger defined governance rules, which produce explainable findings and required responses.
Conceptually: Answer or condition → Governance rule → Finding → Required response. The organisation can see which information triggered a finding, why it matters, what response follows and what evidence supports the underlying facts.
Completing an assessment
- Open the assessment associated with the AI Use.
- Review reused information already captured during registration.
- Answer the factual questions you can reasonably answer.
- Use the question guidance where terminology or consequences are unclear.
- Identify information that needs confirmation rather than guessing.
- Obtain specialist input where required.
- Review and complete the assessment so the applicable deterministic rules can be evaluated.
What happens after assessment?
Depending on the findings, Swell may identify required Controls, Actions, specialist review, Approval, conditions, Monitoring or Review. Completing the questionnaire does not itself approve the AI Use.
Deterministic rules provide consistency and explainability, but they do not replace legal, privacy, security or professional judgement.