ALUD uses artificial intelligence to support analysis and collaboration. We do so with clear limits and human oversight.
How we use AI
- Explain indicators and methodology.
- Summarize authorized documents.
- Suggest questions and possible priorities.
Limits
- AI never presents itself as a CPA, attorney or regulated professional.
- It flags uncertainty and requires human review for high-impact recommendations.
- It respects each organization’s boundaries and role-based access.
Data classification
- Public: information already published or intended for publication.
- Internal: ALUD working material that is not published.
- Confidential: your non-public information. It only enters authorized tools, with training on the data disabled.
- Restricted: personal, health or third-party financial data. Out of scope unless a specific written agreement is signed before work begins.
When a person steps in
Whenever a result an AI system took part in is going to affect an employee, a client of yours or your organization money, or is going to enter a deliverable. You designate the reviewer, their name is written down before we start, and they check where every figure comes from, whether the result contradicts something your organization already knows, whether it is being used for what it was designed for, and who it affects if it is wrong.
Your data
We do not use your confidential information to train third-party models. Data hosted on our platform is isolated by organization and access is role-based. On termination, and at your written request, we return or destroy what we hold, except any copy the law or a backup policy requires us to keep, which remains subject to confidentiality.
The limits of the tool
Artificial intelligence systems produce probabilistic results: they can be incorrect, incomplete or different given equivalent inputs, and that is not fully eliminated by design or by testing. That is why human review is not a formality. We do not guarantee a level of accuracy, a saving or a given return.
When something goes wrong
We log incidents even when no visible harm occurred, and anyone who reports in good faith is not penalized. If an incident affects your information, we notify you without undue delay, with what we know at that moment.
AI usage notice
When AI participates in a recommendation, we maintain human review before important decisions and we do not use sensitive traits to score people.