AI disclosure

Last updated 12 November 2026

How AI is used in WAFL AI, and the commitments we make about it.

What is AI

The website assistant, the voice agent, drafted follow-ups, agents’ messages and the morning briefing are written by an AI model (currently Anthropic’s Claude), using the business’s own information and rules. Voices on calls are synthetic.

Always disclosed

The assistant says it is an AI assistant, the chat window is labelled as one, and it will say so if asked on a call. It hands over to a person for complaints, urgent matters, or whenever someone asks for one, according to the business’s rules.

Grounded in the business’s facts

It may only state facts the business has given it about itself (prices, availability, policies). When it does not know, it says so and passes the question to the team rather than guessing. Where a business enables expert knowledge, it uses recognised independent information and says where it comes from.

Human control

Businesses review every conversation, can correct the assistant’s knowledge at any time, and approve agents’ messages before they are sent unless they choose automatic sending.

Data

Conversations are used only to provide the service to that business, and are not used to train AI models.