AI transparency statement

Where AI is used

AI organises the facts you provide, summarises documents, suggests questions, builds draft timelines and prepares briefs for Kelly. Everything AI-generated is labelled AI in the portal.

What AI is not allowed to do

• Decide whether you have a valid legal claim, or guarantee an outcome.
• Invent legislation, policy clauses or citations — statements are backed by shown sources or flagged as unverified.
• Tell you human help is unnecessary in a high-risk situation, or close your case.
• Send you a formal response in Kelly's name without her review.

Human review

Urgent situations are escalated to Kelly by deterministic safety rules — not just AI judgement. Responses labelled as reviewed record who approved them.

Deadlines

Dates the AI extracts are always labelled potential — verify. We never rely on AI alone to calculate a legal deadline.

Your data and AI providers

Only the minimum relevant excerpts of your case are sent to the AI provider, under settings where your case material is not used for model training.