The AI client promise: When technology supports the work, trust still owns the relationship

AI is entering the client relationship through the back door – in drafts, summaries, reviews, timelines, emails, and internal preparation. Even when clients never see the technology, they feel its effects: faster answers, cleaner documents, better-prepared teams.

But clients do not buy speed. They buy confidence.

For a professional firm, the promise is not simply to use technology. It is to use it in a way that protects trust. That promise becomes real when firms classify AI use before it creates doubt.

The real question is not whether AI was used. It is whether its use changes risk, accountability, or trust.

A useful client promise can be built around four categories: invisible, disclosed, approved, and off-limits. The point is not to disclose everything – that would help no one. The point is to know what belongs in each box.

Invisible: No client communication is needed because the use does not change risk, expose confidential information, or affect accountability.

Examples: Organising internal notes, preparing an agenda, summarising non-confidential material.

Disclosed: The client should be told, in plain language, that AI may support part of the work.

Examples: Document reviews, due diligence, substantive drafts that rely on AI. (Disclosure should not sound like a warning. It should sound like governance: clear safeguards, professional review, and responsibility.)

Approved: The client's express consent is needed when the use of AI involves sensitive information, external tools, regulated data, or matter-specific restrictions. It protects both client and firm before the risk becomes uncomfortable.

Off-limits: Some uses simply should not happen. Confidential files do not go into public tools. Client data is not training material. AI work does not leave the firm without professional review.

No advice is ever defended with: “The system said so.”

I see these scenarios and judgement calls play out every week, in firms across Europe and Latin America. The same use can be treated as routine on one team and forbidden on another. A shared framework turns AI from an individual shortcut into a firm-wide standard.

For firms starting now, the first step is not buying another tool or announcing a grand AI strategy. It is mapping the AI uses already happening: which are simple productivity, which involve client information, which need safeguards, and which should stop. That map can become a short internal guide teams will actually use.

Clients may not choose a firm because it uses AI. They may choose it because its use feels controlled, explainable, and safe.

---

Feliza Bahamonde Wormull is the Managing Partner at SBA Abogados. She studied law at the Universidad Católica, Chile, and at the Universidad Internacional de La Rioja, Spain. She also holds a master’s degree in Digital Law and New Technologies from the University of Salamanca, Spain.